Microsoft's AI Futurist explains how he uses Copilot — and the real-world problems enterprises are solving with agents
Summary
Microsoft's Build 2026 conference highlighted the company's aggressive push into enterprise AI agents, introducing several key platforms and models. Central to this strategy is Microsoft IQ, a new context layer spanning GitHub Copilot, Microsoft Foundry, and Copilot Studio, which includes Work IQ, Fabric IQ, Foundry IQ, and Web IQ for various data and web interactions. The company also unveiled Scout, a personal work agent, and seven new in-house MAI family AI models designed for token efficiency and customization. Marco Casalaina, Microsoft's VP Products, Core AI and AI Futurist, emphasizes the company's commitment to model choice, supporting frontier models from major players alongside its own MAI models. He notes that Microsoft 365 Copilot has surpassed 20 million users, serving as a critical interface for custom agents, and cites examples like Bayer and AEMO using Foundry to deploy agents for tasks ranging from employee assistance to managing electrical grid operations.
Key takeaway
For AI Architects and Directors of AI/ML evaluating enterprise agent deployments, Microsoft's Build 2026 announcements signal a robust, integrated ecosystem. You should consider utilizing Microsoft Foundry for its comprehensive agent management, observability, and rubric-based evaluation capabilities, especially when building custom agents that require secure access to diverse enterprise data via the new IQ services. The ability to publish custom agents directly into Microsoft 365 Copilot or Teams offers a significant advantage for user adoption and workflow integration, streamlining the path from agent development to widespread employee use.
Key insights
Microsoft's agent strategy centers on providing a comprehensive platform for context, governance, identity, and secure data access, alongside diverse model choices.
Principles
- Prioritize model choice and customization for enterprise needs.
- Agents should empower humans by automating grunt work.
- Context, governance, and identity are crucial for agent reliability.
Method
Microsoft's approach involves a layered stack: foundational models (MAI, third-party), hosted agents in Foundry for managed deployment, a Foundry control plane for observability and evaluation, and IQs for agent-facing context.
In practice
- Publish custom agents directly to Microsoft 365 Copilot or Teams.
- Utilize Foundry's control plane for agent cost and performance monitoring.
- Implement rubric-based evaluation for granular agent correctness testing.
Topics
- AI Agents
- Microsoft Foundry
- Microsoft IQ
- Enterprise AI
- Copilot
- MAI Models
Best for: CTO, VP of Engineering/Data, Executive, AI Engineer, Director of AI/ML, AI Architect
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by VentureBeat.