Microsoft AI Update October 2026
John Savill walks through the month’s changes across Microsoft’s AI stack, from models and inference chips to workplace agents and developer tools. He also discusses the organizational, economic, and resilience considerations involved in scaling AI.
Scaling AI in organizations
The opening discussion focuses on four challenges:
- Trustworthy scaling — expanding AI adoption while maintaining trust.
- Organizational redesign — the changes needed to achieve the highest employee productivity gains.
- AI operating economics — a focus on the economics of running AI.
- Sovereignty and resilience — considerations for how AI services are deployed and operated.
Models and inference hardware
The model and hardware portion covers:
- MAI Code 1.1 Flash local.
- The MAI-Base-2 foundation model.
- OpenAI Decisions API.
- GPT-6.1 Terra.
- MAI-Image-2.6 and 2.6-Flash.
- Mistral Large 4.
- Sonnet 5.5.
- The Maia 300 inferencing chip.
This segment begins at 04:56.
Copilot and agent capabilities
Savill reviews updates to the Microsoft Copilot harness, browser automation tooling, and Multi-LoRA adapters. The agent-focused topics also include:
- BYOT SIP for voice agents.
- A customer conversation harness.
- Agent ID support.
- Business process memory.
- Triggers.
The discussion then moves to Copilot Projects, Copilot Workspaces, role plugins, and Copilot for Legal and Finance.
Workplace collaboration topics include referencing previous conversations in Cowork, Teams recap without saving transcription, Work IQ MCP retrieve, and the Insights agent.
These updates begin at 10:03.
GitHub Copilot updates
The developer-tooling segment covers the GitHub Copilot app and Copilot Fusion, followed by:
- Code review Max.
- Copilot Projects.
- Session continuity.
- Copilot Automations.
The GitHub Copilot discussion starts at 20:33.
Infrastructure and integration
The final technical segments cover infrastructure updates, Azure API Management (API-M) integration with A365, and Project Perception updates.