Early insight from our team in San Francisco, combined with our first look at Microsoft’s Book of News 2026.
Our team is in San Francisco for Microsoft Ignite 2026, hearing directly from Microsoft, its partners and organisations already putting its latest technology to work. Microsoft has introduced new capabilities across Microsoft 365 Copilot, Microsoft IQ, Agent 365, Azure, security, data and business applications, and released its Book of News 2026 alongside them. Some of this is generally available; some remains in preview. This is a first-look summary, and we’ll keep testing the detail as more sessions take place.
One message is already clear: Microsoft is moving the conversation beyond individual artificial intelligence (AI) productivity, towards an operating model where people, AI agents, organisational knowledge, business systems, security controls and cloud infrastructure work together.

"The big shift is that AI agents are going from answering to acting, and trust is now what decides who wins. Microsoft Perception is now generally available: AI agents that find and fix threats at machine speed. Windows adds built-in containment and identity for agents. Security is finally the foundation, not the afterthought. Bring on the rest of the week!"
James Pearse, Chief Technology Officer, Intercity
Coming soon · On-demand
Ignite, unpacked
Our team is filming an on-demand episode with the Ignite crew as soon as they're back from San Francisco, going deeper on what they saw and what it means in practice. Sign up and we'll send it straight to your inbox the moment it's ready.
The early message from San Francisco
The opening sessions were framed around “frontier transformation”: AI moving from a tool that helps one person finish a task, to a capability that can take part in how an organisation runs. Three connected shifts stood out:
- Humanisation: Copilot becoming the central place people find information, create work and delegate activity.
- Organisational intelligence: agents working with an organisation’s own context and data through Microsoft IQ, not just a single prompt.
- Trust: stronger identity, permission and risk controls as agents gain real access to information and systems.
Microsoft is describing Copilot as a new operating system for work: AI that can be given an objective, use approved organisational knowledge and business tools, and carry out several stages of work over time. Copilot Home brings different AI experiences into one starting point, with automatic routing to the right mode or model. Much of this remains in preview, but the direction is hard to miss: AI is moving from assistance towards delegated work.
"Microsoft IQ will be a major step towards unlocking organisational intelligence by connecting business data, knowledge and context across Microsoft 365, Fabric and Foundry. The opportunity for organisations is no longer simply about deploying Copilot or introducing AI tools. It's about fundamentally rethinking how work gets done, connecting people, data, applications and intelligent agents to deliver measurable business outcomes!"
David Craft, Microsoft Alliance Director, Intercity

WHAT IT MEANS FOR BUSINESS LEADERS
1. AI moving from assistance to delegated work
Pick a small number of repeatable processes with clear ownership and measurable outcomes. Avoid a vague instruction to "use more AI."
2. Copilot becoming a more central place for work
Factor Copilot into wider digital workplace and adoption planning, not just as a chat tool.
3. AI built around recognisable business roles (sales, service, finance, legal)
Define the outcome first, such as cycle time, quality, cost or risk, before picking the use case.
4. Organisational knowledge becoming a competitive advantage
Check that important knowledge is current, authoritative and correctly permissioned.
5. Employees able to build more of their own tools
Create a controlled route for proposing, testing and scaling ideas, with clear rules on what must be documented and approved first.
6. Agentic commerce changing how customers find products
Review product and service information for accuracy and structure. Treat machine-readable data as part of the customer journey.
7. Customer Zero becoming part of partner strategy
Ask any partner for evidence beyond licence numbers: baselines, outcomes, costs and lessons learned.
8. AI value facing greater scrutiny
Set a cost and usage baseline. Time saved is not automatically the same as cash saved.
9. Governance becoming a leadership responsibility
Require a named business owner for every consequential agent, with a defined purpose and approval points.
10. Trust may matter more than raw model power
Build explainability, human oversight and safe failure into the business case from the start.
WHAT IT MEANS FOR IT LEADERS
Every useful agent brings dependencies: identity, access, data, infrastructure, integration, monitoring and cost.
IT leaders are being asked to support a shift from a small number of AI assistants to a much larger and more varied population of agents. That takes more than enabling licences: it means building identity, access, security, data, operational and financial controls around systems that increasingly act on their own.
1. Agents becoming identities that need managing (Agent 365)
Give every agent a purpose, named owner, approved access profile, monitoring policy and retirement process.
2. Shadow agents becoming a governance issue
Build an inventory covering approved agents, employee-created agents and third-party tools.
3. Agents creating a new security attack surface
Apply least privilege, review connectors and plugins, and require human approval for consequential actions.
4. Model Context Protocol (MCP) connections needing scrutiny
Create an approval and review process for MCP servers and other external connections.
5. Microsoft IQ raising the importance of data architecture
Fix ownership, permissions and metadata. Poor information will produce poor AI outcomes.
6. Copilot-built apps expanding shadow IT
Define a route from prototype to supported production, including security review and documentation.
7. AI operations becoming more connected to business impact
Strengthen service mapping so autonomous operations can prioritise by business impact, not just technical failure.
8. Legacy modernisation becoming part of AI readiness
Include application and infrastructure readiness in AI planning. Copilot cannot make up for a weak underlying estate.
9. Production AI requiring real infrastructure choices
Make deliberate decisions on performance, data residency, isolation and model selection, rather than defaults.
10. AI consumption requiring financial control
Set budgets and attribute agent costs at use-case level, not just overall licence spend.
11. Resilience needing to cover models, connectors and data sources
Test safe failure, human takeover and recovery as part of continuity planning.
12. Incident response needing to account for autonomous activity
Define how an agent can be disabled or contained quickly, and preserve logs for review.
The Intercity view, based on what we know so far
The AI conversation is becoming more practical, but also more demanding. The opportunity is moving beyond buying Copilot licences or experimenting with prompts. Organisations need to understand where AI could create meaningful value, prepare the data and technology beneath it, secure the identities and connections involved, and measure what happens after deployment. The organisations that pair business ambition with IT readiness, rather than chasing every announcement, will be best placed to move from enthusiasm to practical advantage.
Coming soon · On-demand
Ignite, unpacked
Our team is filming an on-demand episode with the Ignite crew as soon as they're back from San Francisco, going deeper on what they saw and what it means in practice. Sign up and we'll send it straight to your inbox the moment it's ready.
Get the full deep dive, head over to Microsoft Ignite 2026 Book of News: news.microsoft.com/events/ignite/2026/book-of-news