AI moves from pilots to production
Microsoft Ignite 2026 returns to San Francisco from November 17 to 20, with an optional pre-day on November 16. After last year’s focus on building a governed agent ecosystem, this year’s event looks set to answer the harder question: how do organisations turn AI agents into dependable, secure, measurable parts of daily operations?
For business leaders, IT teams, developers, and partners, the message is likely to be less about “what AI can do” and more about “how to run it safely at scale.” Microsoft’s agenda already points strongly in that direction, with sessions and labs focused on agentic business applications, agent platforms, data, security, Microsoft IQ, and Frontier Transformation.
At Stellium, we expect Ignite 2026 to mark another step in the shift from isolated Copilot experiments to enterprise AI that is embedded in data, applications, workflows, and governance.
The central theme: Frontier Transformation
Microsoft is framing this year around Frontier Transformation: connecting AI innovation to tangible business outcomes. Rather than treating AI as a separate project, the idea is to embed enterprise context across data, applications, and workflows so AI can support real decisions and actions.
That matters because many organisations have already tested Copilot, built a prototype agent, or explored automation. The challenge now is making those initiatives useful beyond a demo.
A practical example: a finance team may use an agent to review invoices, identify exceptions, check approval status, and prepare a summary for the controller. The value is not the chat interface. The value is that the agent understands the process, uses the right data, follows permissions, and produces an auditable result.
Expect Microsoft to continue positioning this as the path to becoming a Frontier Firm: an organisation where people and agents collaborate in a governed, scalable way, with impact that can be measured.
Agents become operational infrastructure
Last year, Ignite introduced a major shift toward agent governance through Agent 365 and Entra Agent ID. The idea was simple but important: if agents are going to act on behalf of people and systems, they need identities, permissions, monitoring, and lifecycle management.
This year, expect that foundation to mature.
Microsoft’s event agenda includes dedicated themes for Agent Platform and Agentic Business Applications, while the pre-day includes sessions on building a path to production for agents and building a first agent in Copilot Studio. That suggests a focus on practical deployment: designing agents, connecting them to business systems, testing behaviour, managing access, and operating them over time.
The likely questions for enterprises will include:
- Which processes are suitable for agent automation?
- Which data sources and applications should an agent access?
- Who approves an agent’s actions?
- How do you prevent an agent from becoming another unmanaged system?
- How do you measure quality, cost, adoption, and business impact?
These are not just technical questions. They are operating-model questions. Organisations that treat agents as managed digital workers, rather than experimental chatbots, will be better placed to scale.
Copilot and Microsoft 365 get more action-oriented
Copilot will remain central, but expect the conversation to move beyond content generation. Microsoft’s agenda includes innovation sessions on how Copilot, Windows, and Teams connect people and agents to transform work and scale AI confidently.
In practical terms, this could mean Copilot becoming more involved in workflows, as Copilot Cowork already shows, rather than simply helping with documents, emails, or meetings. Instead of asking for a summary, users may ask an agent to prepare a customer review, update a CRM record, create a follow-up task, and notify the right colleague.
For example, a sales manager could ask Copilot to review an account, identify overdue actions, compare recent communications, and prepare a next-step plan. The agent would need access to email, CRM, files, and internal policies, but also clear boundaries around what it can change.
That is where the real value lies: not in generating more content, but in reducing manual coordination and helping people make better decisions faster.
Microsoft IQ: context becomes the differentiator
One of the most interesting areas to watch is Microsoft IQ. Last year, Microsoft introduced the IQ family as a context layer across work, data, and agent behaviour, including Work IQ, Fabric IQ, and Foundry IQ.
This year, Microsoft has made Microsoft IQ a dedicated Ignite content theme and includes a pre-day workshop titled “Build with Microsoft IQ: A hands-on workshop for context-aware AI.”
The underlying idea is important. An AI model is only as useful as the context it can access. A generic assistant can answer a question. A business-aware agent can understand the customer, the document, the approval process, the relevant data, and the next action.
For example, in a customer-service scenario, an agent may need to understand:
- The customer’s history and open cases
- The product or service involved
- Internal policies and entitlements
- Recent communications
- The correct escalation path
Without that context, the agent produces generic answers. With it, the agent can support a meaningful outcome.
Expect Microsoft to show how organisations can bring together structured and unstructured data, business semantics, and work context to create more reliable AI experiences.
Security and governance take centre stage
Security will be one of the most important themes at Ignite 2026. Microsoft’s agenda includes a dedicated Security track, and the pre-day includes sessions on security, endpoint readiness, Microsoft 365 Copilot governance, and preparing the endpoint estate with Entra, Intune, and Windows.
This is a necessary response to the growth of agents. When AI can access systems and take actions, traditional security models need to evolve.
Expect discussion around:
- Agent identity and access management
- Least-privilege permissions
- Data protection and compliance
- Observability and auditability
- Risk detection across AI environments
- Governance for agents built with low-code tools
The pre-day also includes an innovation session on using AI observability to govern agents, uncover risks, and strengthen security at scale. That is a clear signal that Microsoft sees observability as essential, not optional.
For organisations, the practical takeaway is straightforward: before scaling agents, define the controls. Identify sensitive data, restrict access, log activity, test behaviour, and assign clear ownership. Governance should enable adoption, not slow it down.
Azure, data, and infrastructure remain foundational
While agents will dominate the headlines, the underlying cloud and data platform will still matter. Microsoft’s agenda includes themes for Data, Windows, and Frontier Transformation, alongside sessions on how Azure and Windows support mission-critical AI from cloud to edge.
Microsoft has also confirmed NVIDIA’s involvement at Ignite, with sessions and labs covering accelerated computing on Azure, secure agentic AI at scale, physical AI, and scaling intelligence from cloud to edge.
For enterprises, this reinforces an important point: AI outcomes depend on more than a model. They depend on data quality, integration, infrastructure, security, and operational discipline.
A company may have access to powerful models, but if its data is fragmented or its processes are unclear, the AI will struggle to deliver reliable results. Ignite is likely to reinforce the need for a strong data foundation, especially for organisations building agents that act across multiple systems.
What attendees should prioritise
With hundreds of sessions, labs, demos, and partner activities, Ignite can easily become overwhelming. The best approach is to attend with a small number of business outcomes in mind.
For business leaders
Focus on:
- Frontier Transformation and measurable AI outcomes
- Customer stories showing real-world adoption
- Governance, risk, and change management
- How Copilot and agents fit into existing business processes
For IT and security teams
Focus on:
- Agent identity, permissions, and lifecycle management
- Data protection and compliance
- Observability and auditability
- Endpoint, identity, and cloud security
- Windows 365 and secure AI workspaces
For developers and makers
Focus on:
- Copilot Studio and agent development
- Microsoft IQ and context-aware AI
- Azure AI Foundry and multi-model approaches
- Microsoft Fabric and data-rich applications
- Hands-on labs and certification opportunities
Microsoft is offering labs, certifications, expert meetups, theatre sessions, and partner activations, so attendees should reserve time for practical learning rather than only keynote viewing.
What Stellium expects
Our expectation is that Ignite 2026 will continue the shift that began clearly at Ignite 2025: AI is no longer a set of disconnected experiments. It is becoming an operational layer of the enterprise, with agents, context, governance, and security working together.
The organisations that benefit most will not necessarily be those that adopt the most AI features. They will be the ones that connect AI to clear business processes, strong data foundations, and sensible controls.
The opportunity for partners and customers is to move from “Can we use AI?” to “How do we deploy AI safely, measure its value, and scale it across the organisation?”
That is the conversation we expect to see at Moscone Center next week. And it is the conversation that will shape enterprise AI over the next 12 months.