Updates released between June 24, 2025, and July 8, 2025
Early July brought significant developer-focused updates to Microsoft 365 Copilot, emphasizing extensibility, compliance, and unified management. This update cycle particularly benefits organizations building custom agents and IT teams responsible for governance across AI deployments.
Unified Store Arrives in Office Documents
Microsoft introduced the Microsoft 365 unified store directly within Word and PowerPoint, enabling users to discover, acquire, and manage agents without leaving their documents. This in-app store streamlines the agent adoption process—no more context-switching to browse catalogs or rely on IT to provision tools.
Excel support is coming soon, completing the Office suite integration. This unified approach reduces the cognitive overhead of managing separate tool ecosystems for different applications.
User experience transformation: Previously, discovering useful Copilot agents required navigating admin centers or searching through disparate catalogs. Users often didn't know what agents existed or how to access them. The in-app store brings discovery into the flow of work—when drafting a sales proposal in Word, users can immediately browse relevant agents without breaking concentration.
TypeSpec Simplifies Agent Development
Developers can now use TypeSpec as an authoring experience for declarative agents and API plugins in Teams Toolkit. TypeSpec provides a more intuitive, type-safe way to define agent behaviors and integrations compared to traditional approaches.
Additionally, automatic project scaffolding for Graph connector applications within Teams Toolkit generates full production-ready applications from just an API description file, dramatically reducing setup time and complexity.
Developer productivity impact: Building integrations traditionally requires significant boilerplate code, configuration files, and manual setup. TypeSpec and automatic scaffolding eliminate this grunt work, allowing developers to focus on business logic rather than infrastructure. A connector that might take days to build manually now takes hours, accelerating custom integration delivery.
Outlook Email Becomes Agent Knowledge
Makers can now build custom agents that read and reason over Outlook messages, delivering answers that reflect the latest decisions and context stored in email—without data migration or retraining. This addresses a fundamental challenge: email contains critical business context but has historically been inaccessible to AI systems.
Business context preservation: Email threads often contain decision rationale, stakeholder feedback, and historical context that never makes it into formal documentation. When agents can reference this information, they provide more accurate, contextually appropriate recommendations. A pricing agent that can review email negotiations gives better guidance than one limited to price lists and policies.
Compliance and Audit Infrastructure
Microsoft rolled out comprehensive audit and eDiscovery capabilities for Copilot actions and declarative agents through Microsoft Purview. Organizations can now view detailed audit logs and eDiscovery records, simplifying compliance and investigation workflows.
Regulatory readiness: Enterprises in regulated industries need complete audit trails for AI-generated content and decisions. Financial services firms, healthcare organizations, and government contractors must demonstrate what AI systems did, when, and based on what data. These Purview integrations provide the documentation infrastructure necessary for compliance with sector-specific regulations.
Graph Connector Workflow Improvements
Developers using Teams Toolkit to author declarative agents can now select specific Graph connectors to improve knowledge grounding. This granular control ensures agents only access relevant data sources rather than indiscriminately pulling from all available connectors.
The ability to add Dataverse as a knowledge source further expands the types of enterprise data agents can reason over, enabling more comprehensive responses rooted in organizational data.
Copilot Studio Agents Reach Microsoft 365
Organizations can now publish, manage, and use agents built with Copilot Studio directly within the Microsoft 365 Copilot app—across both web and desktop—and in Microsoft Teams. This bridges previously separate ecosystems, allowing Power Platform agents to surface in Microsoft 365 workflows seamlessly.
Strategic integration: Many organizations have invested in Power Platform for process automation and app development. Enabling these investments to extend into Microsoft 365 Copilot creates synergy rather than duplication. An agent built to automate expense approvals in Power Platform can now assist with expense-related questions in Copilot Chat.
Performance Analytics and Agent Management
Copilot Studio gained performance metrics for every knowledge source, showing usage frequency, answer rate, and error rate. This visibility helps organizations spot high-value content and quickly fix low-performing links, keeping agents accurate and helpful.
IT admins received new metadata visibility in the Microsoft 365 admin center for Shared agents, similar to line-of-business application information. This consistency in management experience reduces the learning curve for administering AI agents.
Python Integration in Excel Advances
Excel users can now leverage "Think Deeper" mode when performing advanced analysis with Copilot and Python. This reasoning model integration provides more elaborate planning followed by automatic execution to generate Python code, results, and explanations—improving performance on complex analytical tasks.
Data analysis democratization: Python has become essential for advanced analytics, but learning Python represents a significant barrier for business analysts. Copilot's natural language interface to Python capabilities means Excel users can request sophisticated analyses—regression modeling, clustering, time series forecasting—without writing code.
Unified Microsoft 365 Copilot App Interface
Microsoft updated the Microsoft 365 Copilot app UI to serve as a centralized starting place for AI at work, offering quick access to secure AI chat, search, files, and content creation in one seamless experience. The "Create" module specifically serves as a creative hub for AI-led artifact generation.
Platform Updates Across the Suite
- Purview added Data assessments in Microsoft Purview AI Hub for reviewing sensitivity and access
- OneNote launched Copilot Chat on OneNote for web and Teams
- Teams introduced follow-up question suggestions in meetings
- Viva Connections replaced Feed experience with SharePoint News reader featuring Copilot-powered summaries
- Word provides automatic document summaries on file open, coaching suggestions when rewriting, and contextual prompts on Mac
Agent Center for Centralized Management
The new Agent Center provides a single location to organize, store, and update declarative Copilot agents. Developers can register in-context actions, fine-tune prompts, and test behavior faster, while IT admins can roll out reliable, task-specific Copilot experiences at scale.
Governance at scale: As organizations deploy dozens or hundreds of agents, centralized management becomes essential. Agent Center provides the control plane necessary to maintain quality, security, and consistency across the AI portfolio—similar to how app stores manage mobile applications.
Looking Forward
This update cycle reflects Microsoft's maturation of Copilot from individual productivity tool to enterprise platform. The emphasis on compliance (Purview integration), management (Agent Center), and developer productivity (TypeSpec, scaffolding) signals readiness for large-scale deployments in risk-conscious organizations.
The Outlook email integration particularly stands out as transformative—unlocking email as an agent knowledge source solves a long-standing problem of preserving business context that lives in inboxes rather than formal systems.
For CIOs evaluating Copilot for enterprise-wide deployment, these updates address critical concerns around governance, auditability, and management overhead. The platform is evolving from experimental technology to manageable infrastructure.
With robust developer tools, compliance infrastructure, and unified management, Copilot moves closer to enterprise-grade AI platform status.