Roadmap, security and documentation changes captured for Microsoft Copilot (Microsoft 365).
<p>Users can access Business Applications through Work IQ in Copilot Cowork and use Business Skills to complete end-to-end business processes grounded in their enterprise data and apps. This helps users move from finding the right business application to taking action across key workflows with less context switching.</p>
<p>Business skills are natural-language instructions that capture how your organization gets work done. They represent your business processes, policies, and domain knowledge in a format that agents can understand and follow. Each skill describes how to complete a specific type of work—the steps involved, the information required, and the business rules that apply.</p> <p>Agents discover and use the business skills they have access to as needed at runtime using Work IQ MCP to complete tasks according to your organization's processes. When multiple agents use the same skill, they follow the same process, ensuring consistent behavior across your organization.</p>
<p>Enable agents to execute reusable Work IQ recipes that coordinate supported tools and actions.</p>
<p>Chat naturally with Facilitator without tagging it each time.</p>
<p><strong>Streamline corporate accounting work</strong></p> <p>New accounting capabilities in Excel and Cowork help accountants spend less time on repetitive, manual tasks in the places they already work, including:</p> <p>• <strong>Reconcile accounts with confidence:</strong> Compare balances and transactions, identify discrepancies, and prepare reconciling items for review.</p> <p>• <strong>Prepare journal entries faster:</strong> Draft accruals, reclassifications, allocations, reversals, and recurring entries, with checks for required inputs and balanced entries.</p> <p>• <strong>Turn results into clear reporting:</strong> Analyze financial performance against plan, identify key drivers, and prepare reporting that explains what changed and why.</p> <p>Additional capabilities span audit, tax, FP&A, compliance, and training. Learn more about the full set of accounting capabilities here.</p> <p>Accountants are already putting these capabilities to work. In seven days of limited Insiders availability, more than 8,000 unique users invoked accounting skills over 44,000 times across more than 18,000 Copilot conversations.<br> Across Excel and Cowork, finance professionals will also have the option to start an audit trail whenever they invoke an accounting capability. If selected, this will create a record in their chat history of what Copilot changed and why. This is available today in Excel and coming soon to Cowork.</p>
<p>We are introducing new Microsoft 365 Copilot usage reports for Copilot Search, Copilot Connectors, and Copilot Agents for US Government customers across GCC Moderate, GCC High, and DoD environments.<br> These reports will provide administrators with organization-level and user-level insights to help them understand usage and adoption of these Copilot capabilities.</p>
<p>Educators can use Teach in Microsoft 365 Copilot to create structured unit plans that span multiple lessons or weeks. They can customize the plan components to organize connected lesson plans and quizzes across a unit.</p> <p>#education</p>
<p>Bullseye routing with user groups enables a tiered assignment strategy that automatically expands the pool of eligible representatives as conversation wait time increases. Rather than assigning from a fixed pool or waiting for overflow rules to trigger, this approach progressively widens availability in concentric rings—starting with the best-fit representatives and expanding outward as needed.</p> <p>Administrators configure this scenario by associating user groups with a queue and defining wait time thresholds for each tier. For example:</p> <ul> <li>0–30 seconds: Assign to Tier 1 specialists (primary user group)</li> <li>30–60 seconds: Expand to include Tier 2 representatives (secondary user group)</li> <li>60+ seconds: Include all available representatives (tertiary user group)</li> </ul> <p>Conversation orchestration monitors each conversation's wait time and automatically expands the eligible pool when thresholds are reached. This process ensures that customers are matched with the most qualified available representative, while preventing excessive wait times when the primary pool is at capacity.</p> <p>Key benefits:</p> <ul> <li><strong>Best-fit first</strong>: Prioritizes the most qualified representatives before expanding to broader pools.</li> <li><strong>Automatic expansion</strong>: No manual intervention required—wait time thresholds trigger expansion automatically.</li> <li><strong>Configurable tiers</strong>: Define as many user group tiers as needed, with custom wait time thresholds for each.</li> <li><strong>Seamless integration</strong>: Works alongside other conversation orchestration scenarios like dynamic priority escalation and overflow based on representative availability.</li> </ul>
<p>A new agentic debug run feature will be available in the flow designer. The UI automation repair interface launches while the flow is running, validating each selector used in the flow. It then proposes a new selector in case an existing one does not work as expected, after the user captures the respective element for reference.</p>
<p>The Chat experience in the Copilot app will suggest a better-fit mode based on the user's prompt, if applicable. For example, it will suggest Cowork or other supported modes as they become available. Users will see a suggestion card after they send a prompt in Chat, with an explanation of the better-fit mode and an option for the user to switch to the suggested mode. Their chat conversation will be attached to the prompt in the new mode to maintain context.</p>
<p>App Blocks enables M365 Copilot Chat users to turn their requests and data into rich, interactive experiences directly within the conversation, including calculators, dashboards, forms, planners, visualizations, simulations, comparison tools, and other interactive apps. It helps users move beyond static responses to explore data, adjust inputs, visualize outcomes, and simulate workflows in place, while supporting grounded scenarios where current information can be retrieved and then rendered into an interactive experience.</p>
<p>This feature enables users to easily create new SharePoint sites with Copilot. Users can review and confirm a clear site plan before Copilot builds the site.</p>
<p>Overflow for direct inward dialing leverages conversation orchestration to detect key direct inward dialing-related events—such as the representative being busy, offline, rejecting the call, or failing to answer—and apply administrator-defined fallback actions.</p> <p>Administrators configure natural language playbooks using conversation orchestration at the global or inbound profile level to specify the action to run when a direct inward dialed call can't be attended to or completed. The actions include offering a direct callback to the caller, sending them to voicemail, transferring the call to another queue or external number, or ending the call.</p>
<ul> <li>Define outreach scenario/goal in natural language.</li> <li>Choose desired call-to-action to drive next steps.</li> <li>Provide subject line guidance: fixed suggestion or style recommendation.</li> <li>Set message length and brevity targets.</li> <li>Specify tone rules, including name-usage limits and preferred terminology.</li> <li>Configure brand language and voice guardrails: approved phrases and banned words.</li> <li>Set formatting preferences: bullets, paragraph length, and maximum bullet count.</li> </ul> <p>Business Value:</p> <p>This feature expands the Sales Qualification Agent outreach by letting teams define the outreach purpose, brand-aligned tone (matching both company voice and seller style), the call-to-action, and key messaging preferences proven to resonate with prospects. The result is more relevant, consistent communication that drives higher response rates and, in turn, improves lead qualification rates.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<ul> <li><strong>Segment-specific qualification</strong>: Each agent operates with its own ideal customer profile, knowledge sources, templates, and consent settings, ensuring qualification conversations reflect the nuances of each product, region, or team.</li> <li><strong>Intelligent lead routing</strong>: The system evaluates incoming leads against agent selection criteria and assigns ownership based on configured priority, ensuring predictable and conflict-free lead management. A fallback agent handles leads that don't match any specialized criteria, ensuring no lead goes unprocessed.</li> <li><strong>Exclusive engagement with coordinated handoff</strong>: Only one agent engages a lead at any time. When a lead's interest shifts to another product or segment, the system hands off with full context to the appropriate agent.</li> <li><strong>Shared knowledge foundation</strong>: Agents can leverage common company-level knowledge while maintaining their own specialized sources, ensuring consistent answers across the organization.</li> <li><strong>Supervisor oversight and governance</strong>: A centralized dashboard enables supervisors to monitor agent activity, reassign leads, and track the full qualification journey across all agents.</li> </ul> <p>Business Value:</p> <p>Multi-agent configuration in Sales Qualification Agents enables sales organizations to deploy agents aligned to distinct product lines, regions, or sales teams, each with its own qualification criteria, knowledge sources, and engagement approach. Multi-agent configuration ensures exclusive lead ownership, coordinates handoffs when context shifts, and provides sales managers with visibility across agents. This targeted approach mirrors real-world sales structures, eliminates duplicate outreach, accelerates qualification, and scales coverage without additional seller headcount.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>This feature enables Quality Managers to edit and override evaluation results even after they are marked as completed or expired. While completed evaluations remain read-only by default to preserve integrity, supervisors can initiate an <strong>Edit</strong> action when discrepancies are identified, such as incorrect AI predictions or errors in manual assessments.</p> <p>In edit mode, supervisors can update evaluation responses, adjust scores, and refine comments to reflect an accurate assessment. This ensures that evaluation records represent the final, validated view of agent performance and interaction quality. The capability is available across AI-assisted, AI agent-generated, and manual evaluations, providing a consistent correction experience.</p> <p>To maintain control and governance, only authorized users with supervisory permissions can perform overrides. All changes are tracked through audit signals, including who made the edit and when, and overridden evaluations are clearly flagged. This ensures full transparency for reporting, compliance, and downstream reviews.</p> <p>Business Value:</p> <p>Enables Quality Managers or users with privileges to correct completed evaluations when inaccuracies arise from AI predictions or human review errors. This ensures evaluation outcomes accurately reflect real-world context and business standards. By introducing controlled human-in-the-loop oversight with full auditability, organizations can confidently scale automated and manual evaluations while maintaining trust, transparency, and governance.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<h3>AI agents in Business Central</h3> <p>Business Central introduces AI agents to help users automate repetitive tasks and make business processes more intelligent and efficient. These agents act as specialized assistants that understand context, handle structured data, and collaborate with users to complete end-to-end workflows. For example, the Sales Order Agent simplifies order creation by transforming natural language or email instructions into complete sales documents, while the Payables Agent streamlines vendor invoice processing by extracting key details from incoming invoices and matching them to purchase orders—all reducing manual effort and freeing users to focus on higher-value work.</p> <p>The agent runtime that powers these built-in agents provides core capabilities that enable the use of agentic AI within Business Central. AI agents are defined, configured, activated, and run directly inside the familiar Business Central environment—leveraging the same security model, administration tools, and user experience as the rest of the product. Results are displayed in-context, with intuitive navigation to related data and entities, allowing users to control the agents through the interface they already know. This consistency dramatically reduces the learning curve, builds trust, and ensures efficiency when working with AI agents.</p> <p>The agent runtime ensures that every step taken by an agent is traceable and operates strictly within the access boundaries (permissions) assigned during configuration. It provides a timeline view of all agent activities, allowing users to review what happened and why—enabling a true human-in-the-loop experience. The runtime can also memorize relevant data during execution, use it in subsequent steps, and request additional instructions in natural language when user input is needed. It helps ensure that sensitive operations always require explicit user review and consent, supporting full transparency and control.</p> <p>Business Central business processes are inherently multistep, stateful, and financially consequential. The agent layer must be able to interact with a broad set of ERP capabilities and data, sequence operations with a high degree of accuracy across dependent steps, handle the data respecting permissions, company, environment, and tenant boundaries, provide a contextual audit trail, and keep humans in the loop.</p> <p>Business Central agentic strategy intentionally separates deep, well controlled, in-context and transparent execution within the product from cross-system orchestration. Business Central agents form the foundational layer inside the product, exposing well-defined agent capabilities that in the future will be possible to surface as MCP tools and consume by Declarative Agents in Microsoft Copilot Studio and Microsoft 365. This layered approach enables Business Central-tuned, transparent, managed execution at the ERP core processes, while allowing higher-level Declarative Agents to orchestrate end-to-end business processes across products.</p> <h3>Designing agents</h3> <p>The new in-product experience for designing agents that automate Business Central processes is made available with AI Development Toolkit for Business Central. The experience offers a low-risk environment for envisioning and prototyping new AI agents tailored to your own business scenarios—extending the same agent runtime capabilities used by built-in Business Central agents.</p> <p>Built for rapid iteration, the designing experience lets you create and refine custom agents using natural language instructions, define their behavior, test agent profiles and permissions, and safely experiment with Business Central data in a sandbox environment. This approach empowers partner personas like consultants, product owners, and domain experts, as well as developers and even power users, to explore new scenarios, validate ideas, and build confidence in their agents before moving to production. Whether you aim to automate repetitive business processes, streamline data processing, or deliver proactive insights, the agent design experience gives you the tools to bring your vision to life.</p> <h3>Agent instructions</h3> <p>At the core of every agent are its goals and instructions, defined in natural language. Writing agent instructions is similar to writing prompts for large language models, so you can apply general prompt-writing best practices—and even use tools like Copilot to optimize them. The agent design experience includes a simple editor to help you define, refine, and test these instructions efficiently.</p> <h3>Invoking your agent</h3> <p>You can invoke your agent by manually creating a task for it. When doing so, you can include an additional message that provides specific details or context to complement the agent’s core instructions. Although the ability to design your own agents is in public preview, there aren't yet any built-in integrations to automatically trigger these custom agents based on incoming emails, events, or scheduled recurrences (this can be done by professional developers using the APIs available in the AI Development Toolkit). However, you can simulate these scenarios by including the email header and body in the task message to mimic incoming emails, or by manually triggering the agent to emulate event, UI action, or time-based runs.</p> <h3>Troubleshooting</h3> <p>When running the agent tasks in the designing experience, you can access detailed diagnostic views to help you understand and tune the agent execution. Since an agent’s decisions depend on its instructions, available data, and language model reasoning, you can review execution logs to see its inputs, reasoning process, and outputs. This helps you identify where uncertainty occurred and why specific decisions were made to adjust your agent instructions.</p> <h3>Importing and exporting agents</h3> <p>To make it easy to back up or move custom agents across sandbox environments, the in-product experience allows you to import and export agent definitions. These definitions are stored as simple JSON files containing properties such as agent identity, instructions, assigned profiles, and permissions. We recommend storing these files in a source control system like GitHub to track changes and maintain version history.</p> <h3>Responsible AI</h3> <p>Agents built using the agent design experience adhere to Microsoft’s Responsible AI principles, helping you ensure agent compliance, fairness, and safety.</p> <h3>Billing</h3> <p>As with the built-in Sales Order and Payables agents, custom agents consume Copilot credits as they perform their steps. Credits can be provisioned through prepaid or pay-as-you-go models, providing flexible cost management as your AI usage scales.</p> <p>Business Value:</p> <p>Agents in Business Central are AI-powered team assistants that go beyond traditional automation. Unlike static workflows or hard-coded AL logic, agents can interpret high-level business goals expressed in natural language and translate them into actionable steps. They reason over Business Central data, make informed decisions within defined boundaries, and collaborate with users through human-in-the-loop approvals for critical operations.</p> <p>The benefits are substantial: agents reduce manual effort by automating repetitive or complex tasks, allowing users to focus on higher-value activities. They adapt to dynamic business scenarios that can't always be anticipated during development, offering flexibility that traditional automation lacks. Most importantly, agents operate transparently: every action is logged, permissions are rigorously enforced, and users remain in control at all times.</p> <p>This combination of intelligence, adaptability, and accountability makes agents a cornerstone of the next generation of ERP experiences. With the agent design experience, we're taking the first steps toward enabling you to build your own custom Business Central agents.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Create and enforce policies that prevent risky or noncompliant customer communications.</p> <p>Governance policy checks serve as configurable safeguards that review AI‑generated and autonomous email responses before they are sent to customers. These checks enforce business, legal, and regulatory requirements by validating responses for relevance, prohibited content, and sensitive topics. If a response fails a policy check, delivery is blocked and the violation is clearly identified, enabling supervisors to review issues and take actions to improve outcomes.</p> <p>Governance policies are applied consistently across email engagement channels and seamlessly embedded within agent workflows, ensuring uniform enforcement of quality and compliance standards without requiring manual intervention. This approach empowers organizations to scale AI-assisted and autonomous customer engagement with confidence while upholding trust, safety, and governance.</p> <p>Business Value:</p> <p>Governance policies give you control over AI-driven customer interactions. By setting clear, configurable rules, you make sure every response meets your business, legal, and regulatory standards before it reaches customers. This reduces risk, protects your brand, and keeps communication consistent and trustworthy.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
⏰ Act before July CY2026 · overdue
<p>You can now configure and run multiple Sales Order Agent instances in a single company. Each instance operates independently with its own mailbox, folder settings, email signature, and processing rules.</p> <p>Each agent instance tracks its own KPI metrics, such as emails received, quotes created, orders created, and time saved. The setup page is restructured to support per-instance configuration, and an upgrade path migrates existing single-instance setups to the new architecture. This change allows organizations to set up separate agents for different sales teams, product lines, or customer segments, each monitoring their own personal or shared mailbox.</p> <h3>Improved items search:</h3> <p>When a customer requests items, the agent now uses a multi-layer matching strategy:</p> <ul> <li>Exact identifier match: direct lookup by item number, vendor item number, GTIN, or cross-reference.</li> <li>Semantic attribute matching: extracts attributes from the customer's message (color, material, size) and ranks candidates by relevance using AI.</li> <li>Broader fuzzy search: falls back to a global item search when direct and attribute-based searches produce no results.</li> </ul> <h3>Message and attachment relevance:</h3> <p>The agent now validates incoming messages and attachments against a detailed relevance model before processing. Messages that fall outside the agent's responsibilities (such as offers to sell, requests for discounts, or attempts to extract internal information) are identified and filtered.</p> <p>A separate relevance check runs on attachments, ensuring that non-sales-related documents don't trigger unnecessary processing. The validation considers the full conversation history and the agent's defined scope to make accurate relevance decisions.</p> <h3>Larger attachment files with diagnostics.</h3> <p>The maximum number of pages per PDF attachment has been raised from 10 to 50, addressing customer feedback about the agent not reading larger purchase orders. A new cumulative content length limit of 150,000 characters is also enforced across all attachments in a single email (approximately 50K tokens).</p> <p>When an attachment is skipped, Business Central now stores the specific reason directly on the attachment record (page count exceeded, cumulative content length exceeded, attachment count exceeded, unsupported format, or analyze-attachments not enabled). This information is visible on the <strong>Attachments</strong> page and logged in telemetry, making it straightforward to diagnose why a particular document wasn't processed.</p> <h3>Try it out.</h3> <p>You can now create tasks for the Sales Order Agent without sending an email. From the agent's setup page, a new <strong>Try it out</strong> section lets you compose a message by selecting a sender (from your contact or customer list), entering message text, and attaching files. The agent acts like it received a real email.</p> <p>To help you get started quickly, two sample messages are included. One is a text-only message that references items from your catalog, and the other is a message with a sample purchase order attachment. The sample order is generated based on your actual item data, so the agent has realistic content to work with.</p> <p>These samples are particularly useful for first-time setup, demos, and testing agent behavior with specific scenarios without needing to send an actual email.</p> <h3>Handling of unknown senders</h3> <p>When the agent receives an email from an unregistered email address, you now have three options instead of just one:</p> <ul> <li>Create a new contact: register the sender as a new contact (existing behavior)</li> <li>Use another contact once: pick an existing contact from the contact list and link them to this specific task only, without modifying any master data. The override is stored per task message and does not affect future emails.</li> <li>Update an existing contact's email: select an existing contact and update their email address to the sender's address</li> </ul> <h3>More control over email</h3> <p>New configuration options give administrators more control over how the agent interacts with the monitored mailbox:</p> <ul> <li>Mark email as read: a new toggle to mark processed emails as read in the mailbox. This works in addition to the custom <em>Processed by Sales Order Agent</em> category, giving users a visual signal in their inbox.</li> <li>Mandatory folder selection: the agent now requires you to select a specific mailbox folder before activation. Previously this was optional and could lead to the agent monitoring an unexpected folder. If in doubt, choose <em>Inbox</em>.</li> <li>Stricter activation validation: trying to activate the agent without a mailbox folder now shows a clear error message instead of allowing activation in an ambiguous state.</li> </ul> <h3>Prompts as embedded resources.</h3> <p>The agent's AI prompts (instructions, item search logic, mail template checks, irrelevance validation) are now shipped as embedded resource files within the app extension package. This change reduces latency on first task processing and makes the prompt structure more transparent for engineering iteration.</p> <p>Business Value:</p> <p>The Sales Order Agent now handles more complex organizational structures and higher volumes with less manual intervention. You can run multiple agent instances in one company, each dedicated to a different team or shared mailbox with independent settings and performance tracking. The agent finds the right items more reliably by understanding product attributes, informal descriptions, and identifiers across multiple matching layers.<br> The agent now processes PDFs up to 50 pages (previously, the limit was 10) and clearly shows why any attachment was skipped. It also filters irrelevant messages before processing, so credits aren't consumed by marketing emails or out-of-scope inquiries. If a customer writes from an unregistered email address, you can link an existing contact to that single task without creating duplicates. You can test the agent directly from the setup page without sending a real email, reducing the time to validate your configuration.</p> <p>Enabled for:</p> <p>Enabled for end users automatically</p>
<p>Autonomous email resolution delivers end-to-end handling of customer emails without manual intervention or default case creation. When an email arrives, the system analyzes the request, identifies the underlying intent, and determines whether the information and actions needed to resolve it are available.</p> <p>Using enterprise knowledge, configured business instructions, connectors, and custom agents, the system constructs a resolution dynamically. It generates a response that's context-aware, compliant with business rules, and tailored to the request, then delivers it directly by email—completing the interaction autonomously.</p> <p>If the customer isn't satisfied or the issue can't be resolved automatically, the system creates a case that carries the full conversation context, including customer intent, actions taken, and generated responses. A service representative can then step in without losing context or duplicating work.</p> <p>The capability is delivered through the Case Management Agent. An administrator enables the setting and sets the automation level for each line of business, from fully autonomous resolution to drafting a response for a service representative to review and send.</p> <p>This shifts email handling from a case-first workflow to an intent-driven resolution model, reducing manual effort, improving consistency, and accelerating time to resolution for high-volume customer interactions.</p> <p>Business Value:</p> <p>Autonomous email resolution drives faster, consistent responses and frees service representatives to focus on complex, high-impact issues. It improves customer satisfaction and operational efficiency and scales support without increasing response times and costs by automatically resolving high-volume email inquiries.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>The Customer Insights MCP server allows AI agents and chat experiences to access and understand insights generated by Customer Insights - Data.<br> The Customer Insights MCP server uses the industry-standard MCP architecture to securely share unified profiles and insights with AI agents and chat experiences, providing the data context so AI agents can do their best work.</p> <p>AI tools can use the Customer Insights MCP server to:</p> <ul> <li>Resolve Dataverse contacts and leads to their unified customer profile.</li> <li>Access extended customer information unified by Customer Insights - Data from external sources not available to Dataverse.</li> <li>Understand calculated measures for a customer such as their purchase totals over the last 4 months, or the date of their most recent engagement.</li> <li>Quickly discover what segments the customer is a member of.</li> <li>See if a customer has multiple redundant contacts in your system, and which is considered authoritative for that customer.</li> <li>Get predictive insights for the customer such as Customer Lifetime Value (CLV) and Churn risk, giving you greater insights into the customer's current relationships with your company.</li> </ul> <p><strong>Scenario</strong><br> Sally is a seller who is evaluating a lead named Joe who recently attended a product launch webinar. Sally asks her AI embedded on their Contact and Lead forms if there are any insights about this lead and is delighted to find that the lead Joe is matched to a VIP customer Joseph who has existing paid licenses for several products. Because Customer Insights - Data unifies her company's external Purchase and Rewards data with Dataverse contact and lead data, the agent informs Sally that Joe is a valued Rewards and Loyalty points member, and it is able to provide other AI-generated insights such as the Customer Lifetime Value score, Churn Risk, and the segments Joe is a member of. Armed with a wealth of customer data, Sally is able to engage Joe and upsell Joe on their new product.</p> <p>Business Value:</p> <p>Agents give your marketing, sales, and service teams essential insights and can automate many business processes. But agents are only as good as the data they work with. When you give them access to customer insights that unify your customer data sources, you can get more valuable and actionable insights that enable organizational efficiencies that you couldn't achieve before.</p> <p>Now, more insights generated by Customer Insights - Data can be leveraged from within Microsoft Copilot Studio, allowing you to turn insights into a powerful business asset. Copilot Studio, and even your own custom copilots created in Microsoft Copilot Studio, can now simply and easily get access to more insights, allowing agents to understand, share, analyze, and act on these insights. Every team member gets tailored information and tools to reach their goals efficiently.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>An administrator can enable this feature for the entire organization in the Customer Service admin center. Turning on next best actions populates the Copilot pane with actions that service representatives can use. Important case attributes are consolidated in one place as the representative gathers information through back-and-forth exchanges with the customer. Knowledge articles are suggested so representatives can dig deeper and understand the reasoning behind the recommendations.</p> <p>Business Value:</p> <p>Next Best Action (NBA) allows service representatives to perform a suite of Copilot-powered actions in one click. These actions include, but aren't limited to, drafting emails, updating knowledge articles, updating case notes, and more.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Generative insights for Demand planning in Dynamics 365 Supply Chain Management help you analyze sales history and forecast data across dimensions like product and location.</p> <p>With general availability, generative insights now include a new metric: <em>trends</em>. The system dynamically clusters data based on detected patterns, such as <em>high trending</em>, <em>medium trending</em>, <em>relative flat</em>, and <em>declining trend</em>.</p> <p>You can view aggregated values by selected dimensions or drill down into detailed values for each planning object.</p> <p>Business Value:</p> <p>Generative insights in Demand planning help you identify patterns like seasonality, trends, and signal correlations to improve forecast accuracy. By clustering data dynamically, you can make better decisions, optimize inventory, and reduce costs from overstocking or stockouts. These insights help you meet customer demand, improve service levels, and gain a competitive edge.</p> <p>Enabled for:</p> <p>Enabled for end users automatically</p>
<ul> <li>Customer-first direct callback introduces a configurable, AI-powered experience managed through "callback profiles" in the Copilot Service admin center.</li> <li>Administrators define a callback profile—including greeting message, alternate number capture, callback offering window, retry logic, and AI agent selection—and associate it with queues under overflow handling.</li> <li>For each queue where direct callback is chosen as an overflow action, the administrator also must select a callback profile to ensure a customized callback experience for the queue.</li> </ul> <p>Business Value:</p> <p>Customer-first direct callback gives you a smarter way to manage customer calls. Instead of making customers wait on hold, the system initiates a callback the moment they reach the front of the line, so that they never lose their place. A prebuilt AI agent captures callback details. You can use your own AI agent to run custom workflows like customer verification. Your representatives join the call only when the customer is ready, saving time, reducing missed calls, and boosting productivity.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Build Dataverse solutions by using coding agents with natural language prompts. For example, "I'm building a recruiting system for Zava Construction. I need tables for Positions, Candidates, Interviewers, Interviews, and Feedback - with lookups, a many-to-many relationship between Candidates and Positions, and a self-referential interview chain. Create everything in a Zava Recruiting solution, load sample data, and show me which candidates are currently interviewing."</p> <p>From this prompt, the agent autonomously:</p> <ol> <li>Discovers your Dataverse environment and configures MCP.</li> <li>Creates the solution by using the Microsoft Power Platform CLI.</li> <li>Builds five tables with choice columns, lookups, and a many-to-many relationship.</li> <li>Generates and runs a Python script to bulk-load realistic sample data.</li> <li>Queries across tables to answer your business question.</li> </ol> <p>No skill names. No tool flags. No context switching between docs, CLI, and scripts. The agent handles the orchestration end-to-end.</p> <p>Business Value:</p> <p>By using coding agents, you can eliminate the need to manually create Microsoft Dataverse tables and solutions through the UI. Dataverse skills guide coding agents to connect to your environment and build solutions by using tables, relationships, forms, and views. This process also includes creating sample data.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Power Pages MCP server enables makers to expose site data and operations through a secure, standardized interface compatible with AI clients.</p> <p>In the public preview release, makers can enable Create, Read, Update, Delete (CRUD) operations via the site’s Web API layer without writing custom code. Makers can provision the MCP server endpoint through site settings, allowing customers to securely interact with site data or perform operations using natural language queries through popular AI chat clients.</p> <p>Business Value:</p> <p>Power Pages MCP server empowers you to transform site data and operations into structured tools for AI assistants, enabling natural language interactions for updates and queries. This means faster workflows, less manual effort, and improved user experiences, helping your teams save time and stay productive.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Service Operations Agent in the Copilot Service admin center delivers a unified, conversational experience for configuring Dynamics 365 Contact Center.</p> <p>Administrators use conversational setup and orchestration in Service Operations Agent to streamline setup tasks—configure channels, voice settings, Teams Phone extensibility, and knowledge sources. Through the same single guided assistant, administrators also set up route-to-queue rules and assignment strategies for queues and AI agents—automating and optimizing queue management and agent assignments to drive operational efficiency.</p> <p>With Service Operations Agent, admins can:</p> <ul> <li>Generate and validate configurations in real time.</li> <li>Surface diagnostics and actionable insights.</li> <li>Receive clear explanations for recommendations and setup choices.</li> </ul> <p>By consolidating fragmented configuration workflows, Service Operations Agent reduces errors, shortens deployment timelines, and simplifies onboarding for customers and partners.</p> <p>Business Value:</p> <p>Service Operations Agent is an AI-powered assistant that empowers admins to simplify configuration, speed up onboarding, and ensure continuous validation of changes. By leveraging conversational setup and intelligent automation, the agent eliminates manual complexity, enabling faster deployments, reducing errors, and providing real-time feedback on every change. Designed for enterprise-scale reliability, Service Operations Agent supports seamless transitions, proactive monitoring, and compliance-ready operations.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>Dispatchers often spend considerable time and energy reviewing and adjusting each technician's schedule as conditions change throughout the workday:</p> <ul> <li>A technician has one or more cancellations in the afternoon. To fill the open slots, what requirements optimally keep promise windows, prefer high-priority work, and locate work around the same neighborhoods that the technician is due to travel to?</li> <li>A technician is running over the planned time for the current booking. As a result, they're late for the next booking, and the delay might cascade to subsequent bookings. How do you best adjust the bookings for the remainder of the day?</li> <li>A technician recovers from illness sooner than expected and is now available for work earlier than planned. What's the best way to schedule work for them when they return?</li> <li>A technician’s schedule contains a lot of low-priority work that isn't due for weeks. How can you improve the schedule with work that better meets desired business goals?</li> <li>A technician’s schedule for the day is built manually or with the schedule assistant. How can you rearrange the schedule so that the resource’s travel time is reduced?</li> </ul> <p>In each case, the dispatcher needs to assess the situation, determine which business goals apply, and then update the schedule manually, one booking at a time.</p> <p>With this release of the Scheduling Operations Agent, dispatchers can ask Copilot to provide the best schedule for the affected technician. Within seconds, the dispatcher gets a suggested schedule for review and can apply the updates.</p> <p>You can direct Copilot to optimize for the business goal you want, such as maximizing technician usage or work order priority. Copilot intelligently chooses among existing bookings and new unfulfilled work orders based on factors like skillset match, territories, and customer promise windows. Users can access the feature from the Copilot side pane anywhere within the Dynamics 365 Field Service application.</p> <p>Business Value:</p> <p>The Scheduling Operations Agent for Dynamics 365 Field Service enables dispatchers to quickly and easily optimize schedules for technicians as conditions change throughout the workday, so they can significantly improve their productivity. Dispatchers can ask Copilot to optimize schedules and address issues such as traffic delays, double bookings, or last-minute cancellations that often result in conflicts, gaps, or idle time in a technician's schedule. Instruct Copilot to achieve the desired business objective, such as maximizing technician usage or work order priority while considering factors like skillset match, territories, and promised time windows.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Conversation orchestration in Dynamics 365 Contact Center keeps every conversation actively managed throughout its lifecycle, from initiation through resolution. Instead of applying fixed rules at different stages of a conversation journey, conversation orchestration monitors each conversation as conditions evolve, like wait time increases, fast-filling queues, or service representatives going offline, and responds automatically with the right action. For example, administrators can define the intent using natural-language playbooks as follows: "If a premium customer waits for more than three minutes and no senior agents are available, increase their priority and offer a callback." Conversation orchestration then handles how it's run in real time.</p> <p>You can configure playbooks per scenario through guided templates that keep instructions focused and reliable. Built-in validation catches conflicts before anything goes live, so that you confidently publish changes without needing to test every edge case manually.</p> <p>The key capabilities are as follows:</p> <ul> <li><p><strong>Dynamic priority escalation</strong>: In traditional systems, conversation priority is typically set once, at entry, and never revisited. This means a customer who has been waiting for 10 minutes competes for the same expert as another customer who has just arrived, and a conversation that’s transferred mid-journey resets to whatever the new queue's default priority is. Dynamic priority escalation fixes this by continuously re-evaluating priority as the conversation progresses. You define the conditions such as wait time thresholds, queue transfers, customer tier, or other context and conversation orchestration adjusts priority automatically throughout the conversation lifecycle. High-value or long-waiting customers move forward without anyone watching the queue manually.</p> </li> <li><p><strong>Overflow handling across multiple conditions</strong>: Conversation orchestration supports a broader set of business triggered overflow conditions, allowing organizations to respond intelligently as queue conditions change. Administrators can define actions when a queue is outside operating hours, when all representatives in a queue are signed out or unavailable, when queue wait time exceeds a defined threshold, or when the number of waiting conversations surpasses a specified limit. Depending on the situation, orchestration can automatically offer callbacks, route conversations to alternate queues, expand assignment eligibility, notify supervisors, or deliver contextual customer messages. This ensures customers are never left waiting without a defined path forward. Admins can also configure overflow handling for direct inward dial calls now, when reps are not available or don't accept the calls.</p> </li> <li><p><strong>Assign to a previously engaged or preferred representative</strong>: Customers often need to repeat information when they are connected to a new representative for every interaction. Conversation orchestration helps maintain continuity by reconnecting customers with previously engaged or preferred representatives whenever possible. Assignment decisions take into account historical interactions, queue membership alignment, representative availability, and workload capacity. For asynchronous conversations, organizations can choose to honor representative presence status and active conversation limits before reconnecting customers, ensuring returning conversations are assigned only when the preferred representative has capacity to engage. When the preferred or previously engaged representative is unavailable, conversations seamlessly fall back to the next best qualified representative. This reduces customer effort, improves resolution efficiency, and delivers a more personalized support experience.</p> </li> <li><p><strong>Bullseye routing</strong>: Finding the right representative often requires balancing expertise with response times. Bullseye routing enables organizations to gradually expand the pool of eligible representatives when a conversation isn't assigned within a defined timeframe. Administrators can define progressive routing rings that begin with the most qualified representatives and then expand to broader groups based on skills, proficiency levels, teams, or other business criteria. This approach maximizes the likelihood of connecting customers with the best-qualified representative while ensuring conversations don't wait indefinitely for a specific resource.</p> </li> </ul> <p>These capabilities are the first in a broad and growing set of orchestration scenarios that conversation orchestration will support across the full conversation lifecycle, spanning overflow management, priority handling, assignment strategies, in-conversation messaging, and escalation. Each new capability builds on the same natural-language playbook foundation, so organizations can adopt new scenarios without rebuilding what they have already configured.</p> <p>Business Value:</p> <p>Traditional automatic conversation distribution systems rely on complex rules and static configuration that are difficult to maintain and can't adapt as conditions change throughout the conversation lifecycle. Unified routing with its new AI-powered conversation orchestration uses the power of agentic AI to continuously monitor events throughout the entire conversation lifecycle and responds with the right action when required.</p> <p>This is a shift from a static system to an orchestration layer that remains active and intelligent from the moment a conversation arrives in a queue until it's resolved. The new feature reduces administrative complexity and helps deliver more flexible customer experiences at scale, even under fluctuating demand.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>You simply describe in everyday words what you want to create and the Journey Creation Agent builds the journey for you.</p> <p>With Journey Creation Agent, you can:</p> <ul> <li>Use an AI-powered, guided experience for optimized outcomes.</li> <li>Build in minutes with no technical jargon, complex setup, or cold start. Use natural language to describe the journey you want to create.</li> <li>Drive results–the agent interprets your goal and asks relevant follow-up questions to refine your intent and ensure accuracy.</li> </ul> <p><strong>Scenario</strong></p> <p>A campaign manager wants to re-engage inactive customers but has no time to build a journey from scratch. The manager simply types, “Send a re-engagement email to customers who haven’t purchased in 90 days, follow up with a reminder with a discount.” The journey is instantly generated, with the right audience, content, and structure.</p>
<p>Add the Outreach Optimization Agent to your journey, select the emails you want delivered, and let the agent handle timing and cadence automatically. With Outreach Optimization Agent, you can:</p> <ul> <li><strong>Start with an intelligent baseline</strong>: Outreach Optimization Agent automatically creates a baseline messaging plan for your audience by analyzing overall historical engagement patterns, producing a smart, generic plan that is designed to work well for most customers before any individual optimization kicks in.</li> <li><strong>Listen to incoming interactions and adapt the plan in real time</strong>: The agent continuously monitors incoming signals, including engagement and non-responsiveness, and adjusts follow-up timing and cadence, or exits the plan based on whether the goal is progressing or has been achieved.</li> <li><strong>Respect quiet times</strong>: The agent enforces quiet time settings as a hard constraint, ensuring that messages are never sent outside approved communication windows, even if engagement data suggests otherwise.</li> </ul> <p><strong>Scenario</strong></p> <p>A global retail brand uses Outreach Optimization Agent in a seasonal campaign to re-engage customers. Unlike previous efforts with fixed schedules and repeated reminders, the agent creates a smart baseline plan, then adjusts message timing for each customer based on their engagement. It slows or stops follow-ups when customers show interest and spaces out reminders for unresponsive customers, all while respecting quiet times and consent. This leads to better engagement and less message fatigue, with minimal manual intervention.</p>
<p>Sales Development Agent can turn a prospect’s interest in meeting into a booked seller meeting. The agent asks the prospect to provide preferred dates and times, evaluates those options against the appropriate seller’s calendar, and books the first valid option in the prospect’s stated preference order. The seller and prospect receive the meeting invitation, while the seller receives additional conversation context to support a warm handoff. If scheduling cannot be completed, the agent transfers the request and relevant context to the seller for follow-up.</p>
<p>Semantic model helps Sales Agent understand the structure and meaning of an organization’s CRM data. It creates a connected representation of CRM metadata, including tables or objects, fields, descriptions, data types, relationships, views, pick-list values, and glossary terms. Sales Agent can retrieve this context when responding to questions, improving its ability to interpret business language and answer complex questions that span related CRM records.</p> <p>Dynamics 365 admins can enable and turn on semantic model in Power Apps Maker Portal and configure the data coverage and signals that Sales Agent should use. Admins can configure the underlying data, views, metadata and relationships used in the model.</p> <p>For Salesforce customers, the semantic model will be based on existing Sales agent table and glossary configuration in Sales agent settings or Power Apps Maker Portal. In Power Apps Maker Portal, Salesforce admins will can configure additional Salesforce signals to use in semantic model such as relationships and views.</p>
<p>Autonomous agent that researches, qualifies, engages with leads, scaling sales capacity 24/7.</p>
<p>Give AI agents and Copilot a true understanding of your business data — not just raw schema — so they can answer questions accurately with less manual configuration.</p> <ul> <li><strong>Learns from the systems you already run</strong> — Automatically extracts semantic knowledge from views, forms, and metadata already defined in your Dataverse environment, reducing agent deployment from days of manual setup to near-zero effort.</li> <li><strong>Full visibility for makers</strong> — View generated semantic knowledge per entity — including name, description, and sample questions — directly in the Power Apps maker portal. Search for understanding gaps and enrich missing context with a built-in glossary for domain-specific terms.</li> <li><strong>Governance and privacy built in</strong> — Enable or disable semantic signals at the org level, or at the individual signal-type or table level, giving makers and admins granular control to meet compliance and privacy requirements without sacrificing agent quality.</li> </ul> <p>No manual schema mapping. No custom query logic. No disambiguation between what your data is called and what your business actually means.</p>
<p>Intraday Agent is an autonomous AI agent that helps manage intraday operational scenarios. The following are its high‑level capabilities.</p> <p><strong>Intraday queue tracking</strong></p> <ul> <li>Continuously monitors queue volumes in real time.</li> <li>Detects anomalies when actual volume deviates significantly from forecasted demand.</li> <li>Automatically reforecasts in real time to assess current demand.</li> <li>Notifies supervisors with proactive alerts and recommendations.</li> <li>Automates corrective actions such as adjusting schedules, rescheduling breaks, or reallocating service representatives to maintain SLA performance.</li> <li>Supports customizable instructions to align mitigation actions with business needs.</li> </ul> <p><strong>Intraday adherence tracking</strong></p> <ul> <li>Monitors service representative adherence to scheduled shift activities in real time.</li> <li>Detects deviations such as late logins, extended breaks, or early logouts.</li> <li>Autonomously mitigates deviations through actions such as break rescheduling, shift extensions, overtime activation, or flexible channel reassignment.</li> <li>Supports customizable instructions to align mitigation actions with business needs.</li> </ul>
<p>Multi-grader comparison enables makers to compare the results of multiple graders applied to the same evaluation.</p> <p>The experience highlights agreement, differences, and overall grading behavior.</p> <p><strong>Core capabilities</strong></p> <p>The core capabilities include:</p> <ul> <li>Compare graders side-by-side.</li> <li>Compare scores.</li> <li>Review explanations from each grader.</li> <li>Analyze grading consistency.</li> </ul> <p><strong>Results experience</strong></p> <p>Results include:</p> <ul> <li>Comparative score view.</li> <li>Pass / fail comparison.</li> <li>Grader explanations.</li> <li>Evidence supporting each evaluation.</li> </ul>
<p>Allow agents and agent flows to pause for human input or approval before continuing.</p>
<p>Provide administrators with policy-based controls for governing agents and their capabilities at scale.</p>
<p>Enable applications and agents to access Microsoft Teams meeting transcripts in real time for assistance and insights.</p>
<p>This release expands the optimization capabilities of the Scheduling Operations Agent to include multiple resources. The agent considers the availability of the resources, their current schedules, and unscheduled requirements. It suggests a schedule that honors configured constraints and optimizes it according to the selected goal. Dispatchers can review and compare the proposed schedules with the current schedules before applying changes.</p> <p>New capabilities include:</p> <ul> <li>Optimize schedules for up to 5 resources directly from the schedule board.</li> <li>Optimize schedules for up to 30 resources using an optimization plan.</li> <li>Create custom goals and weight objectives.</li> <li>Create custom optimization scopes to use with your optimization plans.</li> </ul> <p>These capabilities build on the existing public preview of the Scheduling Operations Agent and will be automatically enabled for all organizations that have the agent enabled. Learn more at <a href="https://aka.ms/soa">Scheduling Operations Agent documentation</a>.</p> <p>Business Value:</p> <p>Optimize multiple resources with the Scheduling Operations Agent (SOA). Schedule up to five resources directly from the schedule board or Copilot or perform ad-hoc optimizations for up to 30 resources using an optimization plan. This feature streamlines scheduling, improves efficiency, and supports better resource management.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>The plugin in Copilot Cowork is available out of the box for service representatives. They can use the Customer Service skills after the administrator enables the plugin for them in Cowork settings. The key capabilities are as follows:</p> <ul> <li><strong>Prioritizes and enriches cases</strong>: Surfaces highest-attention cases with suggested next actions.</li> <li><strong>Reviews for criticality and severity</strong>: Flags cases where severity mismatches actual urgency signals.</li> <li><strong>Briefs case at intake</strong>: One-screen customer, history, prior cases, and SLA snapshot.</li> <li><strong>Summarizes root cause and resolution</strong>: Proposes likely cause and ranked next-step resolutions.</li> <li><strong>Matches knowledge articles</strong>: Matches top articles, flags gaps, updates stale content.</li> <li><strong>Drafts replies</strong>: Generates template-based customer email replies for CSR review.</li> <li><strong>Handles escalations and handoffs</strong>: Detects triggers and assembles escalation packet for next owner.</li> </ul> <p>Business Value:</p> <p>The Customer Service plugin in Copilot Cowork helps service teams boost productivity at scale by offloading complex, multi-step case workflows. It ensures every action is grounded in the right Dynamics 365 environment and enriched with Microsoft 365 context, eliminating manual lookup.<br> The custom skills allow teams to tailor Cowork to their workflows, knowledge sources, and triage rules for more precise assistance. The built-in checkpoints provide human-in-the-loop governance, ensuring accuracy, compliance, and trust before any changes are committed.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>As an admin, you can control who can access Copilot Cowork in your organization. The Dynamics 365 Sales plugin is installed by default in Copilot Cowork and available to all users who have access to Cowork.</p> <p>As a seller, you can use the sales skills in Copilot Cowork to:</p> <ul> <li>Understand customer, opportunity, and pipeline context without manually gathering information from multiple systems.</li> <li>Identify changes, risks, engagement patterns, and areas that require attention.</li> <li>Prioritize work across customers, opportunities, leads, and pipeline activities.</li> <li>Generate recommendations and next-best actions based on available business context.</li> <li>Analyze business performance, pipeline health, and sales execution trends.</li> <li>Support decision-making across individual, team, and organizational sales workflows.</li> </ul> <p>Business Value:</p> <p>If you spend the day moving between CRM, email, meetings, calendars, and collaboration tools, Dynamics 365 Sales skills bring that context together in Microsoft 365 Copilot Cowork. By grounding Cowork in Dynamics 365 Sales data and Microsoft 365 signals, you get more relevant insights, recommendations, and actions in the flow of work. This helps you quickly understand customer relationships, opportunity health, pipeline changes, and next steps, so you spend less time gathering information and more time moving deals forward.</p> <p>By combining customer, opportunity, account, lead, activity, and pipeline information with emails, meetings, calendars, and collaboration signals, Cowork can help users understand what has changed, identify what requires attention, and determine appropriate next actions.</p> <p>Enabled for:</p> <p>Enabled for end users automatically</p>
<p>Organizations can securely deploy AI agents to access and interact with Dataverse data by using enterprise‑managed Microsoft Entra agent identities. This feature enables agents to be provisioned as Dataverse agent users with dedicated, least‑privileged security roles based on their intended capabilities. This feature ensures agents have only the access required to perform tasks autonomously or on behalf of users.</p> <p>By using identity‑based role assignment and full auditing of agent‑driven actions, organizations can enforce security and compliance policies across both human users and AI agents while maintaining traceability of data access and changes within Dataverse environments.</p> <p>Business Value:</p> <p>This feature enables secure, compliant management of AI agents in Microsoft Dataverse, which reduces risk and improves governance for autonomous and interactive agents.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>The Copilot Cowork plugin for finance and operations apps enables users to move from business intent to insight to execution in a single experience. Copilot orchestrates work across ERP data, emails, documents, spreadsheets, collaboration tools, and downstream business processes.</p> <h3>Key capabilities</h3> <ul> <li>Orchestrate workflows across ERP and Microsoft 365 data sources.</li> <li>Analyze business data and generate contextual insights.</li> <li>Reason across finance, operations, and supply chain domains.</li> <li>Generate recommendations and prepare next steps.</li> <li>Execute actions while keeping humans in the loop.</li> </ul> <h3>What this means for customers</h3> <p>The value is not just accessing ERP data through AI. Copilot Cowork enables agentic workflows where it can reason over business context, gather relevant information, evaluate options, and generate recommended actions.</p> <p>This reduces manual coordination across systems, accelerates decision-making, and improves consistency in how business processes are executed.</p> <h3>User experience shift</h3> <p>Users can interact with ERP through business intent rather than navigating complex application layers and disconnected tools. This creates a more natural way to work, where users focus on outcomes instead of system interactions.</p> <h3>Preview scope</h3> <ul> <li>Initial set of scenarios across finance and supply chain.</li> <li>Limited action coverage for transactional workflows.</li> <li>Available to selected preview customers.</li> </ul> <p>Business Value:</p> <p>Enable users to move from business intent to insight to execution faster by orchestrating work across ERP data, emails, documents, and collaboration tools. Reduce manual coordination, accelerate decision-making, improve workflow consistency, and shift user time from information gathering to applying business judgment.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>Microsoft 365 Copilot Chat is becoming a consistent way that users interact with AI across Microsoft experiences. With this update, that same chat experience is available in Dynamics 365 Customer Service, with service-aware grounding and workflows.</p> <p>Service users can start in Microsoft 365 Copilot Chat and then:</p> <ul> <li>See and prioritize work (for example, top cases and what needs attention).</li> <li>Draft customer communications and prepare next steps with Microsoft 365 context when relevant.</li> <li>Identify trends and operational signals across cases to help supervisors and leads intervene earlier.</li> <li>Discover answers from support knowledge.</li> </ul> <p>Business Value:</p> <p>Bring the familiar Microsoft 365 Copilot Chat experience directly into Dynamics 365 Customer Service, so supervisors, leads, and customer service representatives can get answers, summarize work, and take next steps using both Dynamics 365 data and Microsoft 365 context, without switching tools.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>As a seller, you can:</p> <ul> <li><strong>Ask questions in context</strong>: Ask about opportunity research from Sales agent.</li> <li><strong>Start concise, go deeper on demand</strong>: Begin with action-oriented insights, then request deeper analysis only when needed.</li> </ul> <p>Business Value:</p> <p>You need guidance that adapts to the moment you’re in a meeting preparation, a stage move, or a specific risk you want to validate. With the ability to ask about opportunity research from inside Sales agent, you get right-sized, action-first insights. You can quickly act on what matters now, then go deeper only when you choose. This reduces time spent hunting for answers, improves seller confidence at critical stages, and meet sellers where they are in their flow of work.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>This feature enhances Dynamics 365 Project Operations by extending its native work breakdown structure effort tracking to include approved time captured in Field Service when the Field Service and Project Operations integration is enabled and time flows in at the project task level.</p> <p>Approved Field Service time now contributes directly to task-level WBS execution metrics in Project Operations when time is captured against a project task. Task metrics such as effort completed, effort remaining, percent complete, estimate at completion, and schedule variance are calculated using approved field labor under the same native project rules used for time entered directly in Project Operations.</p> <p>Because these calculations occur at the leaf task level, their impact rolls up automatically to summary tasks and the project root. Effort tracking views, reprojections, and project-level status indicators reflect real execution in the field, rather than inferred progress or manual reconciliation. Project managers can assess delivery health, forecast outcomes, and identify schedule risk using execution metrics grounded in approved field work.</p> <p>This enhancement doesn't introduce new execution logic or formulas. It enables existing Project Operations WBS effort tracking behavior to include Field Service execution when time is captured against project tasks.</p> <p>This feature is only available when the Field Service and Project Operations integration is enabled.</p> <p>Business Value:</p> <p>When field technicians complete work tied to a project task, you see that progress reflected immediately in your project health metrics. Approved Field Service time flows directly into task-level work breakdown structure (WBS) tracking in Dynamics 365 Project Operations, so effort completed, percent complete, estimate at completion (EAC), and schedule variance stay accurate without manual reconciliation.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>End users of Microsoft 365 Copilot will see improved Microsoft Dataverse, Dynamics 365, and Power Apps record references in Context IQ with richer, more descriptive labels that include environment, app, and table information. This makes it easier to identify and select the correct records during Copilot interactions.</p> <p>Business Value:</p> <p>Enables Microsoft 365 Copilot to surface business data with meaningful context, reducing confusion from generic labels and helping users quickly identify the right records to drive faster workflows and more reliable decision-making.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>With email classification, administrators can define a custom taxonomy of categories and subcategories to automatically classify inbound emails in Dynamics 365 Customer Service. AI analyzes the email subject and body, then stores the assigned category or subcategory on the email activity record. Case Management Agent, automatic record creation rules, Power Automate flows, reporting, and other downstream processes can use this value to determine how to handle the email.</p> <p>Administrators can enable email classification and define category and subcategory names and descriptions. They can also set conditions that determine which emails are classified, such as specific mailboxes, sender domains, languages, and so forth. When the feature is enabled, the system qualifies inbound emails and automatically classifies them by using the Case Management Agent flow. Administrators can also manually update an assigned category or subcategory, or disable classification to stop processing new emails.</p> <p>Business Value:</p> <p>Email classification gives you control over how incoming emails are managed before any action is taken. It automatically sorts messages so you can ignore irrelevant ones, let the Case Management Agent handle routine issues, and focus on those that truly need human attention. This reduces clutter from non-support emails like thank-you notes, newsletters, and spam, allowing both AI and your team to concentrate on what matters most. The result is time saved, lower processing costs, and the ability to scale automated email handling confidently without adding operational risk.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>MCP for Analytics provides AI agents with structured access to ERP analytics data and tools. It ensures agents use the same definitions and models as users.</p> <p>Key benefits:</p> <ul> <li><strong>Data Access</strong>: Gives agents access to ERP analytics data for analysis and insights.</li> <li><strong>Metrics and Dimensions</strong>: Provides access to key metrics, dimensions, and relationships used in reporting.</li> <li><strong>Semantic Model</strong>: Exposes the full semantic model to support consistent interpretation of data.</li> <li><strong>Tool Access</strong>: Enables agents to connect with tools like Power BI and Business Performance Analytics.</li> </ul> <p>MCP makes ERP analytics data agent-ready, reliable, and explainable.</p> <p>Business Value:</p> <p>MCP for Analytics allows agents to access ERP analytics data and tools with context. It ensures consistency across metrics, dimensions, and hierarchies so agents can reason reliably. This improves trust in insights, reduces duplication, and simplifies development of explainable agent experiences.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>This capability enhances queue handling by intelligently detecting two key scenarios where conversations would otherwise remain unserved:</p> <ul> <li><p><strong>Queue transitions to out-of-operating hours</strong>: The system continuously monitors active conversations that entered a queue before closing time. If the queue transitions to out-of-operating hours while conversations are still waiting, it automatically applies admin-defined overflow actions. Customers can be guided to leave a voicemail, request a callback, or be transferred to another available queue, ensuring a smooth end-of-day experience without manual intervention.</p> </li> <li><p><strong>No service representatives are signed in</strong>: The system evaluates real-time representative availability. If all representatives assigned to a queue are signed out, it immediately triggers the configured fallback action instead of waiting for in-queue overflow timers. This is especially valuable in scenarios such as mass logout events, unexpected breaks, or holidays not captured in operating hours.</p> </li> </ul> <p>Administrators configure both scenarios using natural language prompts in conversation orchestration, enabling fine-grained control over queue behavior. This ensures customers are always routed to the best possible outcome, even when direct support is unavailable.</p> <p>Business Value:</p> <p>This capability ensures customers are never left waiting when representatives aren't available. Whether a queue transitions to be out-of-operating hours or all support representatives are logged out, the system proactively routes conversations to configured fallback options such as voicemail, callback, or alternative queues.</p> <p>Organizations benefit from reduced wait times and lower abandonment, improved customer satisfaction during peak load or unexpected staffing gaps, and a more predictable service experience regardless of representative availability. At the same time, this eliminates the need for manual queue draining or extended shifts, helping protect service representative wellbeing while maintaining consistent service levels.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>With the dynamic tools of the Dynamics 365 ERP MCP server, large language models often have multiple options for tool-calling sequences to perform an operation or retrieve data. The orchestration can select the tool execution sequence and can include more tool calls than are required. This enhancement to the MCP server improves performance and accuracy of the server response by enabling the server to use optimal tool execution sequences to perform operations with a single point of entry for MCP tools and a semantic index of the available menu items, entities, and actions.</p> <p>Business Value:</p> <p>The Dynamics 365 ERP MCP server now uses a semantic index of target ERP menu items, entities, and actions to improve performance and accuracy of tool calls by optimizing the sequence of tool executions for agentic operations.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>By combining forecast results with contextual insights, planners can gain a detailed understanding of the key factors influencing each forecast rather than simply viewing the numbers. This increased transparency builds confidence and supports more informed planning decisions.</p> <p>Key benefits include:</p> <ul> <li>See symmetric mean absolute percentage error (SMAPE) values.</li> <li>View historical data and forecast results in one place.</li> <li>Understand forecast drivers through AI-generated explanations.</li> <li>Gain greater confidence in planning outcomes.</li> <li>Make faster and more informed business decisions.</li> </ul> <p>These capabilities help planners spend less time analyzing forecast changes and more time acting.</p> <p>Business Value:</p> <p>Get a clear view of historical data alongside a calculated forecast supported by AI-generated explanations. This feature enhances planners' understanding of each forecast, builds confidence in the results, and enables better planning decisions.</p> <p>Enabled for:</p> <p>Enabled for end users automatically</p>
<p>Assign to a previously engaged or preferred expert leverages historical interaction context, queue membership alignment, and expert availability, to ensure customers are reconnected with experts who already understand their issue. When the previous or preferred expert is unavailable, assignments seamlessly fall back to the next best qualified expert, preserving service levels and reliability. This capability reduces repeat explanations, accelerates resolution times, and improves customer satisfaction, helping businesses lower cost to serve while delivering more consistent, personalized support experiences at scale. Administrators define assignment logic using natural language-based playbooks in conversation orchestration, with configurable lookback period to determine the time range until when the system should look for a previously engaged expert.</p> <p>For asynchronous conversations, administrators can also choose to reconnect customers with the same expert by checking the expert’s current presence status and honoring active conversation limits, ensuring returning conversations are assigned only when the expert has capacity to engage.</p> <p>Business Value:</p> <p>By automatically assigning customers to a previous or preferred expert, you deliver faster, more personalized support. By routing repeat or follow up conversations to the most recent or preferred expert within configurable time windows, you reduce customer repetition, improve continuity, and accelerate resolution. When the previous expert is unavailable, the system intelligently falls back to the next best qualified expert, ensuring service levels are maintained. This leads to lower handling times, higher customer satisfaction, and improved expert efficiency.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>This feature includes the following functionalities:</p> <ul> <li>Automated customer payment and remittance advice import, scanning, and mapping using AI processing in the integrated Power Platform environment.</li> <li>Matching remittance advice data with data in Dynamics 365 Finance.</li> <li>Seamless integration of remittance advice details into Dynamics 365 Finance customer payment journals and the settlement process.</li> <li>Reports that provide insights into payment and settlement statuses, outstanding invoices, and overall cash flow.</li> </ul> <p>Business Value:</p> <p>The ability to import remittance advice streamlines the accounts receivable process by enabling businesses to efficiently manage and reconcile payments from customers. By automating the import, scanning, and consumption of remittance advice data, businesses save valuable time and resources. This capability improves cash flow management and financial reporting.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p><strong>Agentic simulations</strong> are an AI-driven capability of Service Operations Agent that enable organizations to evaluate how their contact center behaves under realistic customer conditions before those conditions occur in production.</p> <p>Administrators define simulation scenarios using natural-language prompts. Agentic simulations then generate realistic synthetic conversations, runs them end-to-end through the configured contact center, and calculates the contact center KPIs.</p> <p><strong>What problems it helps solve</strong></p> <ul> <li><p><strong>Readiness of self-service experiences</strong>: Assess how AI agents handle real customer intents, variations in behavior, and edge cases.</p> </li> <li><p><strong>Effectiveness of assisted service flows</strong>: Validate routing, assignment logic, and service representative experience in human-assisted scenarios.</p> </li> <li><p><strong>Confidence in scale and load readiness</strong>: Understand how configurations and operations perform under realistic volume and fix any reliability and latency issues.</p> </li> <li><p><strong>Preparedness of service representatives</strong>: Enable safe practice and evaluation using realistic customer interactions without impacting live customers.</p> </li> </ul> <p><strong>Key capabilities</strong></p> <ul> <li>Natural-language prompt interface for defining simulation scenarios</li> <li>Simulation of varied customer intents, personas, and behaviors</li> <li>End-to-end execution across AI agents, routing, and configurations</li> <li>Out-of-the-box KPI dashboards and scorecards</li> <li>Reusable and saved prompts, including canned scenarios</li> <li>Safe, pre-production execution with no impact on live customers</li> </ul> <p>Agentic simulations are currently available for the voice channel in the English language only. They support flows without Dual-Tone Multi-Frequency (DTMF)-based menu interactions and up to five conversations in a simulation.</p> <p>Business Value:</p> <p>Every contact center decision, such as deploying an AI agent, adjusting routing, onboarding new staff, or preparing for higher volumes, ultimately answers the same question:<br> “Will this work as expected when real customers arrive?”</p> <p>Agentic simulations provide a clear, data-backed answer before changes reach production.</p> <p>By generating realistic customer conversations and running them end-to-end through the contact center, organizations can see how their self-service, assisted service, and operational setup behave under real-world conditions. Instead of relying on assumptions, limited testing, or post-go-live corrections, teams gain measurable insight into experience quality, readiness, and scale.</p> <p>This feature transforms how contact centers operate:</p> <ul> <li>Experience quality improves because AI agents and service representatives are evaluated against realistic scenarios.</li> <li>Operational risk decreases because configuration and routing issues are identified early.</li> <li>Time to value accelerates because onboarding, training, and change validation happen faster.</li> <li>Scale becomes predictable because load behavior is understood before demand increases.</li> </ul> <p>Agentic simulations turn uncertainty into confidence, helping organizations ship changes faster, train teams better, and deliver consistent customer experiences at scale.</p> <p>Enabled for:</p> <p>Enabled for admins/makers/marketers automatically</p>
<p>As a sales leader, you can configure a Sales Development agent once, then share or clone its configuration across your organization. Distribute the agent’s playbook, guidelines, guardrails, and settings to other business units to provide a consistent experience across sales teams.</p> <p>The feature supports these capabilities:</p> <ul> <li><strong>Export a configuration</strong>: In Teams, ask the source agent to package its configuration into a file. For example, enter Export your configuration.</li> <li><strong>Import a configuration</strong>: Attach the exported file to a new agent, then ask the agent to apply the configuration.</li> <li><strong>Create a one-time copy</strong>: The copied configuration is a snapshot, not a synchronized connection. Changes made later to the source agent aren’t automatically applied to the target agent.</li> </ul> <p>Knowledge sources, such as uploaded documents, FAQs, and URLs, aren’t included in the exported configuration. Add them directly to the target agent after importing the configuration.</p> <p>Business Value:</p> <p>Scaling your sales operations just got easier. Instead of manually configuring each Sales Development agent from scratch, you can now set up an agent once with your organization's best practices and guardrails, then share or clone it across your entire team. This means faster deployment, consistent outreach to prospects, and more time for your sellers to build relationships instead of repetitive work. Whether you're expanding to new markets or rolling out agents across multiple business units, you'll maintain quality and compliance while reducing the time it takes to get your agent up and running.</p> <p>Enabled for:</p> <p>Enabled by admins/makers/marketers for end users</p>
<p>When Copilot Chat uses memory to personalize a response, users can now see what informed the answer. Select Sources beneath the response, then open the Personalization tab to view the custom instructions, saved memories, memories learned from conversations, and past chats that shaped the response. From there, users can return to a cited chat to pick up where they left off or review and delete saved memories. Memory citations appear only when memory is turned on.</p>
<p>Provide a unified intelligence layer that helps agents discover, reason over, and use organizational knowledge.</p>
<p>Enable agents to use supported REST endpoints as tools through governed Work IQ integrations.</p>
<p>Enable A2A and MCP tasks to request missing information or authorization and then resume.</p>
<p>Enable durable long-running A2A and MCP tasks with progress tracking, later result retrieval, and cancellation.</p>
<p>Enable agents to retrieve relevant Microsoft 365 content and context through Work IQ.</p>
<p><strong>Sales Development Agent now integrates with Salesforce, enabling organizations to use SDA with their existing Salesforce CRM.</strong> SDA can access leads, manage prospect records, log sales activities, and keep engagement actions—including qualification, disqualification, and ownership changes—connected to the CRM workflow. The integration uses Microsoft’s Salesforce Model Context Protocol (MCP) server to connect SDA with Salesforce.</p>
<p>Work IQ enables agents to create and edit Microsoft Office documents using organizational context, helping users produce and update content more efficiently.</p>
<p>Users can now schedule appointments and meetings directly from chat. Simply ask Copilot to create an appointment or schedule a meeting with colleagues, and it will find available times, book rooms, draft agendas, and send invites—all within the chat experience.</p>
<p>Copilot in OneNote can generate new content and directly edit when in edit mode. In this mode you can create, rename, and delete pages, sections, or notebooks as well as add or edit content in pages. Copilot can bring content from across WorkIQ as well as complete actions across other workflows such as creating Word, Excel, and PowerPoint files.</p>
<p>We're rolling out new capabilities for Copilot Notebook audio overview that make it easier to navigate longer content with topic-based episodes, new audio formats such as Critique and Debate, support enterprise information protection with sensitivity labels, and expand audio overview context to more content types.</p>
<p>Users will be able to access models that require usage-based billing (UBB) from the model selector in Chat. Tenant admins will need to allocate credits so that users can interact with these models. If users consume their allotted credits, they will be able to request credits from their tenant admins.</p>
<p>Business Skills are natural language instructions authored and stored in Microsoft Dataverse that teach Sales Agent how work gets done within a specific organization. Business Skills enable organizations to create natural language instructions for Sales Agent to follow, which represent their business processes, policies, and business knowledge. Business skills can represent common sales workflows such as lead qualification, pipeline reviews, account planning, opportunity scoring, account brief creation, or organization-specific sales methodologies. Sales Agent uses the instructions in Business Skills to deliver responses that align with the customer's business processes and guidance.</p> <p>Sellers can interact with Sales agent using natural language and receive responses that are guided by their organization's Business Skills, following admin defined sales processes, applying company specific guidance, and surface insights that reflect how the organization operations.</p> <p>Administrators enable the required Dataverse capabilities, create and manage Business skills within Dataverse, and control Business skill access. Business skills can be authored using natural language instructions and associated with the organization's CRM environment.</p>
<p>Commercial Journeys helps information workers stay on top of their most important work when browsing in Microsoft Edge. Powered by AI, it proactively surfaces unfinished, new, and recurring tasks on the Copilot new tab page- with context-rich cards that explain why each task matters now. One click opens a guided execution flow that delivers ready-to-use outputs (emails, documents, presentations, and more), so users can complete their unfinished work faster with higher quality. This feature will be available to users with a Microsoft 365 Copilot Premium license in Edge.</p>
<p>Incoming Teams calls don't wait for a break in your day. Intelligent call delegation provides AI-powered calling assistance so you can focus on the conversations that matter most. It answers calls and gathers context, helping you decide whether to pick up and engage the caller. It can also schedule follow-up appointments through an integration with Microsoft Bookings so you can reconnect with callers at a more convenient time.</p>
<p>Expand Context IQ so users can discover and select additional agents, skills, and content sources while composing prompts.</p>
<p>Enable Work IQ to invoke custom agents and agents that call supported third-party tools.</p>
<p>Legal plugins in Copilot provide purpose-built skills to help legal professionals summarize, draft, review, and redline legal documents. Available in Word and Cowork, these plugins can use work context from Microsoft 365 and connect to supported third-party legal sources.</p>
<p>Discover easier ways to use Copilot in OneDrive<br> The new Prompt Gallery offers ready-to-use prompts for common file tasks, helping you get started quickly, discover what Copilot can do, and customize prompts to fit your needs.</p> <p>Access the Prompt Gallery from:</p> <p>The Copilot button in the bottom-right corner of OneDrive<br> The + button in the Copilot chat pane<br> Select Open Prompt Gallery to explore prompts across categories such as Create, Analyze, Find, and Collaborate.</p>
<p>Copilot brings agentic video creation to Clipchamp Video Hub. Users can generate, edit, and refine videos from prompts—grounded in explicit documents or implicit WorkIQ knowledge. Preview and playback video directly in chat, then continue iterating towards a finished video.</p>
<p>Upload skill (.md file) in Brand Kit and use within PPT Copilot</p>
⏰ Act before October CY2026 · overdue
<p>Federated Copilot connectors (MCP type) currently let Microsoft 365 Copilot read data from third-party services in real time. With this update, the connector tools that create, update, and delete content in those services are unblocked in Copilot Chat, so users can complete work without leaving Copilot. Copilot acts with the user's own account and permissions at the third-party service, and every create, update, or delete action requires the user's explicit confirmation before it runs. Admins can view each connector's read and write/delete tools in the Microsoft 365 admin center and disable connectors that don't meet their organization's policies.</p>
<p>Local inferencing expands Microsoft Copilot’s sovereign controls by enabling AI inferencing for supported Copilot interactions to occur within the applicable local geography.</p> <p>Launching in Australia, India, the United Arab Emirates, the United Kingdom, and the United States, this capability gives organizations greater control over where AI inferencing occurs, helping them address sovereignty and regional requirements. Inferencing is the processing step when an AI model executes a prompt to produce an output or response.</p>
<p>Copilot Cowork makes it easy to delegate work. Describe the outcome you want and Cowork automatically grounds the work in your emails, meetings, messages, files, and data.</p>
<p>Edit with Copilot in Word is now available to users in Office 365 GCC, enabling government customers to work with Copilot as a co-creator directly in your document to create, refine, and transform documents more efficiently.</p>
<p>Copilot Chat can now better match search results from text that appears inside images embedded in Word and PowerPoint files is user's working files — such as scanned contracts in Word, text from tables, text from images — content that was previously difficult to search and find.</p>
<p>The new Meeting Shared Files feature allows users to add relevant files and links to their meeting for easy access and to automatically share that content with all meeting participants. AI in the meeting, like the Facilitator agent, uses these files for targeted grounding to answer questions and enhance skill functionality.</p>
<p>Copilot Chat will soon be available across chats, channels, calls, and meetings, helping users stay informed and catch up quickly.</p>
<p>Copilot Notebooks now lets users add Power BI reports as references, bringing report data directly into their notebooks alongside files. Copilot uses this information to help generate more relevant summaries, presentations, and briefs grounded in the organization's data.</p>
M365 Copilot now supports Writing blocks. Edit drafts, memos, emails, and more inline while using Chat for easy editing and iteration.
Self-serve sync connectors enable users to securely connect their own external data sources, such as Jira and Confluence Cloud, to Microsoft 365 Copilot. Using their own credentials and permissions, users can synchronize content they already have access to, making it discoverable in Copilot Chat and Microsoft Search. Administrators maintain oversight through tenant-level controls, including staged rollout, visibility management, and connector disablement.
M365 Copilot now supports code blocks, which lets you preview code, charts, diagrams, and more directly inline while using Chat. Previously, users could preview code side-by-side in a code canvas.
<p>This release follows the availability of organization prompts, providing admins the ability to upload custom prompts for their organization in the Copilot app home screen. Now, admins can delegate organization prompt publishing to users, AAD security groups, and distribution groups. These delegated users and groups will be able to create, edit and delete organization prompts at the tenant level through the Prompt Lab, without having to access the Microsoft 365 Admin Center. Admins can continue to monitor and provide oversight through the current experience in the Microsoft 365 Admin Center.</p>
<p>Federated Copilot connectors enable users to securely connect Microsoft Copilot to their favorite third party sources, retrieving data in real-time using the Model Context Protocol (MCP). Federated Copilot connectors do not store or index customer data in Microsoft - access happens in real time using the user’s own identity, while administrators retain full governance and control through the Microsoft 365 Admin Center. With GA, federated Copilot connectors will be supported across Researcher agent and Microsoft 365 Chat, allowing users to bring data into their workflows.</p>
With improved Copilot file skills in OneDrive, you can go from files to finished work directly from chat. Copilot helps you find the right content, analyze files and data, create polished outputs like summaries, dashboards, and presentations, and collaborate by saving or sharing the results from OneDrive.
Users will be able to insert Agents and Skills in prompts for specialized tasks. They can add them from the Agents and Skills tab after selecting 'Add work content' from the plus (+) menu. Users can also access Agents and Skills by typing / and @ in their prompt.
<p>Copilot Notebooks now supports JPG and PNG files as knowledge sources, allowing users to bring images directly into their projects alongside traditional documents. Copilot can use the content within these images, including text, charts, diagrams, and other visuals, as grounding context to help generate more relevant presentations, reports, and briefs.</p>
<p>Copilot Notebooks now supports CSV and TSV files as knowledge sources, allowing users to bring structured lists and tables directly into their projects alongside traditional documents. This additional context helps Copilot generate more relevant presentations, reports, and briefs.</p>
<p>Create HTML with Copilot in SharePoint or upload your own HTML from another source. Once stored in the pages library, you can render HTML as a page in SharePoint, make additional edits, and share with your audience. HTML gives you new ways to present your content beautifully alongside our existing ASPX pages.</p>
Spend less time catching up on call recordings to resolve customer issues faster. For recorded call queue calls, the Queues app now supports intelligent call recaps for recorded calls that includes AI-generated summary notes, key discussion points, and follow-up actions. Navigate to the call record in the Queues app and select the Recap tab to access.
<p>After a recorded call is completed, use Copilot in the Queues app to quickly catch up on key details, uncover insights, and resolve customer issues faster without having to manually search through a transcript or listen back to a recording. Navigate to the unified call history in the Queues app, select the recorded call record, and launch the Copilot Chat sidecar from within the Recap tab. You can then ask Copilot to generate a summary or answer questions about specific details related to the call.</p>
Copilot memory within Researcher brings forward relevant context from your past interactions—helping ground insights, personalize outputs, and reduce repetitive setup as you explore complex topics.
<p>AI-powered Status Report in Planner Agent turns your plan data into a polished, stakeholder-ready update in minutes. Just ask Planner Agent for a status report and it pulls your task data and transforms it into a clear narrative covering overall status, progress, risks, and what's next. You can easily customize the report to your needs by specifying the audience, tone, length, and sections you want.</p>
<p>Surface new model-driven, dynamically placed Answer Cards across top 7 segments: weather, sports, finance, images, video, places and news for Entra & MSA users across all surfaces.</p>
<p>Workforce Insights (WFI) Agent in Microsoft 365 Copilot is an agent designed to help leaders, managers, and their delegates quickly understand their organization by turning complex workforce data into clear, actionable insights. With WFI Agent, leaders can quickly explore their organization’s structure, team composition, skills inventories, and flexible work patterns to move from insight to action in a single, unified experience directly in Copilot or Teams.</p>
<p>We've introduced a new experience for Copilot in the apps—a consistent entry point across apps that sits above your work and understands the context beneath it. Rather than scattering touchpoints across the interface, it anchors Copilot as one connected system across Microsoft 365, surfacing relevant actions that help you stay in flow. With this release, we are rolling out this new experience - already available across other Outlook endpoints, as well as in Word, Excel and PowerPoint - to classic Outlook.</p>
<p>Multimodal capture enables users to transcribe audio, take images, and type notes in a single session. Copilot then creates a structured Copilot Page with insights and user-captured content, saved to a user-selected Copilot Notebook. Designed for capturing offline work moments—including in-person conversations, whiteboard sessions, and personal notes. This capability is already generally available on OneNote for iPhone and iPad.</p>
This release enables M365 Copilot to reason over Power BI enterprise data and returns grounded answers directly from Power BI reports and semantic models. Users can ask natural‑language questions and receive accurate answers.
<p>Agents that are published to Agents 365 are already managed in the Microsoft 365 Admin Center. If the agent implements the Agent2Agent protocol, it can be used as a connected agent for the agents that users build in Agent Builder or Copilot Studio.</p>
<p>Session and Response Sharing let Microsoft 365 Copilot users share Copilot chat content through a link. Session Sharing shares a full chat session, while Response Sharing shares a single Copilot response. The shared content opens as a read-only view, and recipients can continue from it in their own Copilot chat. This gives Copilot users an easier and more controlled way to share useful Copilot outputs with coworkers instead of copying and pasting. It helps teams reuse work, share context, and collaborate more efficiently while keeping sharing within organizational controls.</p>
<p>Introducing built-in Planner Agent chat with natural language Q&A, smart task discovery, and in-plan task management.</p>
After meeting recap is automatically generated in the translation language chosen during the meeting, users can continue to change the language of the meeting recap after its generated.
Express mode ensures agents and apps receive timely and reliable responses by dramatically accelerating flow run times. This reduces flow failures, improves user experience, and eliminates costly interruptions caused by timeout issues. Organizations benefit from more dependable automations, higher adoption, and smoother operational workflows.
Copilot Notebooks in the Microsoft 365 Copilot app (Android) now lets users organize related chats, output creations, and references into a persistent AI workspace. Copilot uses the Notebook's accumulated context to ground responses, so work continues across sessions instead of starting over each time.
<p>This feature enables IT admins to customize a portion of the user notification messages for recording and transcription during Teams calls.</p>
This feature saves you time by automatically connecting your meetings to the correct CRM records. You have the right deal insights at your fingertips when preparing for customer conversations, so you can focus less on updating records and more on selling with confidence and context.
A multi‑channel outreach capability, spanning web chat, voice, SMS, and email, is critical for meeting customers in the channel they naturally prefer and increasing the likelihood of timely engagement. By activating the agent across multiple surfaces, we ensure continuity of conversations, reduce drop‑off when a single channel fails, and support the FY27 goal of always‑on, end‑to‑end engagement from initial intent to close. This creates higher response rates, faster qualification, and a more seamless handoff to sales motions.
<p>You’ll soon be able to bring your Microsoft Planner into Microsoft Cowork. In Cowork, you can view, create, and update your plans, buckets, goals, and tasks without switching apps. Cowork goes a step further by helping you execute your Planner tasks across Microsoft 365, such as drafting and sending emails, scheduling meetings, posting updates to Teams, and working with files. It can then write progress back to Planner with your approval. You can move from planning your work to getting it done in one place.</p>
Unlock the full value of your structured data by bringing SharePoint lists directly into your Microsoft Copilot Studio agents. With native support, customers can ground AI responses in up-to-date, business-critical data—such as tasks, inventory, customers, or operations—without manual data movement or complex integrations. This enables more accurate, context-aware copilots that reflect how your business runs, helping teams make faster decisions, reduce errors, and automate workflows with confidence.
<p>We will be introducing a new Usage report covering Copilot Chat. This report will include total active users, Copilot Chat usage with a breakout per Microsoft 365 app, and user-level activity insights (anonymized by default). #copilotcontrolsystem</p>
We will be introducing a new Usage report covering Microsoft 365 Copilot. This report will include total enabled and active users, Copilot usage with a breakout per Microsoft 365 app, and user-level activity insights (anonymized by default). #copilotcontrolsystem
By blocking the use of maker's credentials by AI agents, admins can increase the security and compliance of agents in their organizations: - Prevent unauthorized access to sensitive systems or data that the maker's credentials might have access to. - Help organizations comply with data protection regulations (such as GDPR or HIPAA) by ensuring only authorized identities are used for data access. - Enable IT and security teams to enforce organizational policies around identity and access management. - Increase trust among stakeholders and IT departments, making it easier to adopt AI agents at scale.
<p>The agent node lets you embed reasoning directly into your workflow. This means fewer manual handoffs, faster processing times, and workflows that can handle nuanced decisions without additional development effort, helping teams automate more of their end-to-end processes and reduce operational overhead.</p>
This feature helps organizations reduce security risk by preventing agents and flows from being shared when they rely on unsafe identities, such as maker credentials. By detecting identity oversharing early and enforcing safe‑sharing policies at publish and share time, customers can avoid identity leakage, privilege escalation, and unintended access - while still allowing makers to build and iterate safely within approved guardrails.
<p>Ask Copilot about an image in Edge enables users to instantly analyze images from the web by sending them directly to Copilot from the Edge context menu, making it easy to ask questions, gain insights, and understand visual content without leaving their browsing experience. Admins can control availability to this feature using the EdgeEntraCopilotPageContext and Microsoft365CopilotChatIconEnabled policies.</p>
<p>Deeper insights into Cowork adoption and impact accessible from the new AI Cost dashboard and Copilot dashboard. Includes export capability. All metrics can be queried by analysts with the new AI cost query in advanced insights.</p>
<p>Copilot Notebooks in OneNote now suggests artifacts to create, such as Word documents, Excel workbooks, and PowerPoint presentations, based on Microsoft 365 knowledge and the content in your notebook. Select the suggestion and Copilot will generate the artifact.</p>
<p>Save AI-filters generated by the Planner Agent for future use on your plans.</p>
<p>Copilot in PowerPoint helps you create, edit, and refine presentations through natural language—right inside your deck. Start a new presentation or build on an existing one by asking Copilot to generate slides, update content, improve layouts, and polish design, all while preserving your formatting, structure, and branding. Copilot can also connect to your brand kit to apply branded templates, insert brand-approved images, and check for brand compliance. Available to Microsoft 365 Copilot (Premium) users.</p>
<p>Search, chat, and explore the web from one search box with Copilot-suggested actions and curated work content to help your users get things done faster on the Copilot new tab page. Users will see upgraded Copilot new tab page functionality and features.</p>
AI spend and usage insights for GitHub Copilot will be available on Viva Insights for managers with at least 5 direct reports (scoped access), Global analysts and Microsoft Global administrators. They can track spend and make data-driven decisions for AI spend and usage in GitHub Copilot. The dashboard is enabled by default to ensure immediate access.
Copilot Answers in Copilot Search (in the M365 Copilot App on web and desktop) will provide more concise responses and more prominently highlight the ability to continue in the conversation in Copilot chat on the right-hand sidebar.
Insights into AI spend (Copilot credit usage) at a group or team level will be available in Viva Insights as a dashboard and in Advanced insights. The services in scope of this release are Cowork and Work IQ API. The dashboard experience is available to managers with at least 5 direct reports (scoped access), the Insights analysts, and Global administrators and is enabled by default to ensure immediate access to usage-based cost insights. For users to be able to use Cowork or WorkIQ API, your tenant must have usage-based billing setup via Cost Management in the Microsoft 365 admin center. Leader access to tenant-wide data is out of scope for this release. The same insights will be available for Insights Analysts and Global Administrators in Advanced Insights.
<p>Capture on Windows helps you bring conversations and ideas from meetings, brainstorming sessions, and everyday work into Copilot Notebooks. Capture audio, images, and notes in one experience, then let Copilot turn them into structured notes with key insights, decisions, and action items, in a Copilot Notebook of your choice</p>
A new admin control in the People Skills settings lets administrators permanently remove their organization's People Skills data from the tenant. With this new control, organizations can delete their entire skills library along with all users confirmed, inferred, and imported skills data—giving them full control to offboard from People Skills in support of data governance and compliance needs. Once deletion is initiated, skills will no longer appear on user profile cards or be used across Microsoft 365 skills-based experiences such as Copilot and Viva. The control is admin-only with no automatic changes.
Outlook Emails as References in Copilot Notebooks bring email content into the notebook experience, enabling users to ground Copilot in the conversations, decisions, and context that drive their work. By adding emails as knowledge sources, users can connect context from their emails directly to their projects, enriching notebooks beyond files alone and improving downstream outputs like presentations and briefs. Available in Copilot Notebooks across Microsoft 365.
<p>Users can select from multiple voice style options for their voice experience in M365 Copilot.</p>
<p>Enables metered, pay‑as‑you‑go access to Work IQ endpoints, allowing developers to invoke agents and capabilities without requiring pre‑assigned licenses, with usage billed based on the agents and models invoked per request.</p>
Copilot Suggested Rename is an AI-powered OneDrive feature that automatically recommends clear, descriptive file names based on a file’s content—helping users organize and find their files more easily with minimal effort. What the feature does: - Analyzes the content of a file to generate context-aware filename suggestions - Presents 3 recommended names directly within the rename dialog - Allows users to pick a suggestion with a single click Where it shows up: - Within the rename dialog for supported files - Within the post-upload toast notification when a single supported file is uploaded Supported file types: - Microsoft Office documents Word (DOCX) PowerPoint (PPTX) Excel (XLSX) - Other document formats PDF Markdown (MD) - Images
Feature description: Model Context Protocol (MCP) based agents in Microsoft 365 Copilot Chat will now be able to surface rich, interactive UI widgets directly within chat. This enhancement enables developers to deliver more engaging and structured agent interactions. Users will experience these widgets when interacting with agents that implement them, and admins will continue to manage these agents through the Microsoft 365 Admin Center.
<p>We’re introducing support for Model Context Protocol (MCP)-based agents in Microsoft 365 Copilot . This update enables developers to build Copilot agents using MCP servers and allows users to interact with these agents using familiar experiences. This change supports extensibility and improves agent capabilities for organizations.</p>
Declarative Agents with Actions enable Microsoft 365 Copilot to take real-world actions by invoking APIs, plugins, and enterprise workflows directly within the chat experience. Using custom actions, declarative agents can call external systems, execute operations, and retrieve or update data, moving beyond Q&A to full task completion. These actions are powered through API plugins, MCP tools, and enterprise connectors, allowing seamless integration with line-of-business applications and services. The Copilot platform dynamically reasons over available actions and selects the appropriate tool to fulfill user intent, enabling multi-step workflows such as creating records, submitting requests, or triggering business processes within a single conversational flow
<p>Let people search and query Dataverse business data from Microsoft 365 Copilot.</p>
Allows developers to invoke the Researcher agent through the Work IQ endpoint to perform multi‑step analysis, deep reasoning, and synthesis across enterprise content as part of agentic workflows.
<p>Copilot People Connector now supports importing people data with access controls into Microsoft 365. This release enables organizations to bring people data from external systems into Microsoft 365 along with associated access control lists (ACLs), ensuring Copilot respects existing permissions defined in source systems. The imported data can be scoped to specific individuals and groups, allowing Copilot to reason over people data while honoring access restrictions.</p>
<p>Integrate proprietary systems, dynamic knowledge sources, and custom actions into your agentic workflows with MCP support, accelerating extensibility while preserving workflow reliability. This integration improves reuse across agents and workflows, reduces duplication of integrations, and embraces an emerging open and extensible agent ecosystem.</p>
<p>Unlock deeper insights into Copilot agent adoption with flexible, self-serve analytics. Insights global and partition analysts can access and combine granular agent usage metrics with agent and organizational attributes to uncover agent adoption trends, export and customize reporting, and drive more informed decisions, without relying on predefined dashboards.</p>
We're introducing dynamic tool discovery for declarative agents in Microsoft 365 Copilot. Developers can now add, update, or retire tools on their MCP servers without republishing the agent - so end users get the latest agent capabilities immediately, with no deployment cycle in between. Support for federated Copilot connectors will be a fast follow to declarative agents at a later stage. Dynamic tool discovery is backed by Microsoft's enterprise RAI and security protections, and customers retain full control over agent deployment and availability to users and groups.
<p>Copilot in PowerPoint lets you create, edit, and refine presentations through natural conversation, directly in your presentation. You can start a new presentation or build on an existing one, asking Copilot to generate slides, update content, improve layouts, and polish design—while preserving formatting, structure, and branding. Copilot uses your files, meetings, emails, and more to help shape content and iterate quickly, and it connects to your brand kit so you can apply branded templates, insert brand approved images, and check for brand compliance. Available to Microsoft 365 Copilot–licensed users.</p>
<p>Key Microsoft 365 Copilot metrics in custom person queries and reports will have values up to 2-3 days old from the current date. This will provide fresher copilot usage analysis to analysts in Viva Advanced analysis.</p>
Agent Builder now supports using a SharePoint list as a knowledge source for agents in Microsoft 365 Copilot. Makers can ground an agent in structured list data so users can ask questions and get responses based on information stored in a selected SharePoint list. For this release, each agent can reference one SharePoint list with up to 20,000 items, alongside other knowledge source types. Support for list attachments and lookup columns is not yet available.
Vision in Microsoft 365 Copilot helps you make faster, more informed decisions by understanding context from what you share on screen or camera. Copilot can analyze visual inputs from shared desktop screens and provide detailed insights about the content. You can also use vision on mobile during a real-time voice session to share your device camera, front or back, and ask questions about what is in front of you. Copilot analyzes the live camera feed or desktop screen in real time and provides spoken, context-aware answers grounded in work and web data.
<p>A day level Copilot adoption Power BI report reflecting the most recent copilot activity & freshest insights.</p>
<p>Copilot in Forms is being enabled for Microsoft Forms in Government Cloud environments, bringing end-to-end AI assistance to forms, from creating a form, to refining and sending to analyzing results. This rollout includes key Copilot features such as: - Draft with Copilot - Distribution and monitoring - Copilot Insights - On-canvas Copilot features such as Rewrite question, Continue writing, Answer explanations, and more.</p>
Introduces a new REST endpoint for Work IQ that serves as a unified entry point for invoking agents and workflows, providing a consistent, scalable alternative to legacy Copilot Chat endpoints.
Supports connecting to remote MCP servers through the Work IQ endpoint, enabling agents to securely invoke external tools and services beyond the local runtime for richer, distributed agent orchestration.
Enables programmatic access to first‑party and tenant‑defined declarative agents via the Work IQ endpoint, allowing applications to trigger specialized agents for domain‑specific tasks within broader workflows.
A refreshed design for the Microsoft 365 Copilot web and desktop app simplifies the user experience to drive engagement with Copilot. This includes updates to the chat home screen, chat response layout, and navigation pane.
Authoritative Sites feature empowers administrators to designate specific SharePoint sites as official, trusted sources of information. By classifying these sites as authoritative these high-quality, credible content, such as company news, policies, and updates, are prioritized across Copilot Chat and Copilot Search experiences
<p>We are introducing a new Power Automate action that enables automated document generation using document generation (DocGen) forms in SharePoint.<br> Generate document from form (Preview) allows customers to create documents from a template by mapping data inputs to predefined template fields.<br> Documents are generated as part of an automated flow and stored in SharePoint.<br> This capability supports scenarios such as creating contracts, invoices, letters, or reports using structured templates and form-driven inputs.<br> The action works by mapping input data to template fields and returns the generated document along with file metadata such as location and status.</p>
Meetings as References in Copilot Notebooks bring Teams meeting content into the notebook experience, enabling users to ground Copilot on the discussions, decisions, and context that drive their work. By adding meetings as knowledge sources, users can connect transcripts, notes, chats, and shared content directly to their projects, enriching notebooks beyond files alone and improving downstream outputs like presentations and briefs. Available in Copilot Notebooks across Microsoft 365.
The Microsoft365CopilotChatIconEnabled policy is becoming the standard for configuring Copilot Chat. Extension and sidebar policies will no longer affect the appearance or functionality of Copilot Chat.
Today, Copilot can show up to 25 meeting results, which can limit visibility in larger calendars. We’re expanding this limit so you can see more relevant meetings at once making it easier to find, review, and act on the full context you need.
<p>In addition to capturing real-time notes and meeting outcomes during scheduled or hybrid meetings, the Facilitator agent assists you with notes, decisions and actions for in-person meetings in Teams Rooms on Android. Simply invite Facilitator with one tap of the room console. Meeting notes appear on the front of room display or touch board, are available via meeting recap when shared, and deleted if no one selected. Rest assured no data remains in the room. Available with Teams Rooms Pro.</p>
<p>Understand who is building a habit using Copilot with the new power user insights in Copilot dashboard</p>
You can now create a presentation with Copilot in PowerPoint straight from the PowerPoint Home in file menu.
<p>Users can export de-identified, row-level agent metrics and attributes from the Agent Dashboard. Exported data provides detailed insights into agent usage and user-level agent interactions across the organization, with personal identifiers removed.<br> #copilotcontrolsystem</p>
You can now create a presentation with Copilot in PowerPoint just as you open the PowerPoint web app.
Introducing built-in Planner Agent chat with natural language Q&A, smart task discovery, and in-plan task management.
Copilot has been enhanced to interpret and ground responses using images embedded in files such as Word documents (.docx), PowerPoint presentations (.pptx), and PDFs. This capability allows Copilot to extract insights from visual elements like charts, diagrams, and screenshots, ensuring that answers are not only text-based but also informed by graphical context. By combining textual and visual understanding, Copilot delivers more comprehensive and contextually accurate responses for complex queries
Get answers to queries quickly with Copilot's concise AI summaries at the top of Outlook search results. Users can dig deeper by opening Copilot chat directly from the Copilot summary card.
<p>In addition to capturing real-time notes and meeting outcomes during scheduled or hybrid meetings, the Facilitator agent assists you with notes, decisions and actions for in-person meetings in Teams Rooms on Windows. Simply invite Facilitator with one tap of the room console. Meeting notes appear on the front of room display or touch board, are available via meeting recap when shared, and deleted if no one selected. Rest assured no data remains in the room. Available with Teams Rooms Pro.</p>
Currently admins can set Purview DLP policies to protect sensitive data from being sent to any Generative AI apps, users get blocked from sending their prompts to LLM when this is triggered and stops their workflow. This new add-on to the feature shows a new UI that will give users the option to be redirected to M365 Copilot which will open and navigate to this M365 Copilot tab and allow them to send the same prompt there.
<p>Copilot Chat can now better match search results from Scanned PDFs on Copilot users' files they are working on — content that was previously difficult to search and find.</p> <p>This feature will work for new/updated user's working files on a go-forward basis.</p>
You can now choose your preferred image model including OpenAI’s GPT Image, Flux, or Auto for smart, dynamic selection when creating or editing images with Copilot in PowerPoint, or when generating a new presentation or page.
Copilot will be able to find meetings based on topics or keywords from relevant data like meeting body, chat and transcription.
<p>This feature empowers authors to create rich, interactive charts with AI in SharePoint when editing SharePoint pages. Authors can describe the chart they want to create in plain language within a chat on a SharePoint page.</p>
<p>Teams will store AI-generated meeting memory files for meetings where Intelligent Recap was turned on, allowing Copilot (and Facilitator) to retain key insights from meetings beyond transcript retention periods and when a transcript was not saved. This enables Copilot to continue answering questions about past meetings even after the transcript expires or the meeting is over. Admins can also configure how long this AI-generated meeting memory file is stored.</p>
Customers can now view and understand their agent landscape through the new Overview page that will be included as part of the Agent Dashboard in Microsoft Viva. This experience will provide customers with key at-a-glance insights on adoption in their organizations and visibility into who’s creating agents used in their organization.
Enable Global and Insights admins to turn on additional attributes as filters in Power BI reports, including reserved attributes and your organization’s custom attributes. Users who have access to Power BI reports can then use these newly enabled filters to tailor reports to their business needs.
Recording and transcript owners can now grant recording and transcript access to specified people when copying or sharing recap link, and the specified people can access the recap without having to request access.
<p>Facilitator now automatically detects open questions raised during a Teams meeting and offers to help answer them using web search. When a participant asks a question (e.g., “What is an LLM?”) and no one answers, Facilitator posts a prompt offering to find an answer. Participants can select “Yes” to have the agent retrieve and share the information.</p>
Copilot Notebooks allows users to generate Excel spreadsheets using the content and references gathered in a notebook. Copilot drafts a structured spreadsheet grounded in the notebook’s context, which can be opened and edited in Excel, helping users move from ideas and source material to a ready‑to‑use spreadsheet with fewer manual steps.
Copilot now provides coaching feedback on email drafts via chat and can apply suggestions directly to your message in Outlook. Users have the option to choose which suggestions are applied, and Copilot updates your email in place.
Responses are generated by GPT and Claude working together, with GPT drafting an initial response and Claude reviewing for increased response accuracy, completeness, and citation integrity.
Council in Researcher runs multiple leading AI models from different providers in parallel in one place, so you can quickly compare outputs and see consensus vs. conflict) to make high-stakes decisions with greater confidence—without switching tools.
The Copilot Dashboard adoption landing page will be updated to show a unified view of Copilot adoption across the organization. Instead of the default page showing only on Microsoft 365 Copilot (licensed) usage, the page will include all Copilot user adoption, covering both M365 Copilot and Copilot Chat usage that does not require a license. A License filter will allow users to switch between All, M365 Copilot (licensed), and Copilot Chat (unlicensed) views.
<p>Intelligent recap can now be generated without retaining a transcript after the meeting. Meeting organizers can choose this setting from the meeting options before the meeting, or organizers and eligible participants can enable it during the meeting. After the meeting, the AI Summary will be available, but other elements of the recap may not be available (Chapters, Topics, Mentions, Audio recap).</p>
Microsoft is announcing the General Availability (GA) of federated Copilot connectors for Microsoft 365 Copilot. Federated Copilot connectors enable users to securely connect Copilot to popular third‑party sources and retrieve data in real time using the Model Context Protocol (MCP). These connectors do not store or index customer data in Microsoft services; access occurs in real time using the user’s identity, while administrators retain full governance and control through the Microsoft 365 admin center. With GA, federated Copilot connectors will be supported in Researcher agent, Microsoft 365 Chat, and Agent Mode in Excel, enabling users to bring external data directly into their workflows.
We are introducing a refreshed chat-first design in the Microsoft 365 Copilot mobile app, bringing a cleaner, more intuitive interface to users. This update includes a modernized design language, streamlined navigation, and other improvements that make it easier and faster for users to start, continue, and manage chats on the mobile app.
Multimodal capture enables users to transcribe audio, take images, and type notes in a single session. Copilot then creates a structured Copilot Page with insights and user-captured content, saved to a user-selected Copilot Notebook. Designed for capturing offline work moments—including in-person conversations, whiteboard sessions, and personal notes. Other endpoints coming soon.
Mind Maps in Copilot Notebooks are interactive visual artifacts that help users understand notebook content through a grounded map of key topics, themes, and relationships. Users can explore the Mind Map, view summaries for individual nodes, and use Notebook chat to get more detail about a node. Available across OneNote and Microsoft 365 Copilot App.
Create an interactive report, visualization, or dashboard using Copilot Pages. Copilot uses web data and the latest LLM models and can also ground responses in your Work data.
Copilot Notebooks enables users to generate PowerPoint presentations directly from the content and references collected in a notebook. Using the notebook’s curated context, Copilot creates a structured, editable slide deck that users can open and refine in PowerPoint, helping turn research, notes, and insights into shareable presentations more quickly.
Copilot Notebooks allows users to generate Word documents using the content and references gathered in a notebook. Copilot drafts a structured document grounded in the notebook’s context, which can be opened and edited in Word, helping users move from ideas and source material to a ready‑to‑use document with fewer manual steps.
Use intelligent summaries in Copilot Dashboard to quickly surface what’s working and where targeted attention can accelerate Copilot adoption. Intelligent summaries highlight key adoption trends to focus on areas of success. Suggested prompts enable deeper exploration of underlying trends and drivers.
Spoken language detection is now fully automatic. Teams will automatically detect each speaker’s spoken language and update it in real time as the conversation evolves. Manual spoken language selection will no longer be available. This applies to both live captions and transcripts when Interpreter is enabled or when multilingual speech recognition is turned on in meeting options, helping deliver more accurate language recognition and a more consistent multilingual meeting experience.
Intelligent meeting recap will now include video-based recaps. Video recap creates narrated video highlights from recorded meetings, featuring key takeaways and short clips that showcase important moments.
You can choose to select Anthropic models, in addition to OpenAI models, when editing a Word document using Copilot.
When create or edit a presentation with Copilot in PowerPoint, you can now select which model to use, including both Open AI and Anthropic models. Available to Microsoft 365 Copilot–licensed users.
In Outlook, Copilot now works side by side with users to draft, refine, and format emails conversationally, powered by the same agentic capabilities available in Word, Excel, and PowerPoint.
Give users a fast, built‑in way to capture screenshots and include them in Copilot prompts, helping them communicate visual context more easily and receive more accurate, actionable assistance.