Copilot: Inside Agents, Sora 2 & Updates

Key insights

• Microsoft Ignite 2025 video overview: the presenter demos the newest Copilot features and agents announced at Ignite.
  It shows practical examples so viewers can see how the tools work in real scenarios.
• Agent Mode inside Office apps: agents now run directly in Excel, Word, and PowerPoint.
  They can analyze spreadsheets, build formulas, draft documents, and automate slide layouts using natural language prompts.
• App Builder Agent and workflow creation: the video walks through building a simple app workflow with the new agent.
  This feature helps non-developers prototype and connect services faster while keeping steps traceable.
• Sora 2 and AI video creation: Copilot Create now includes Sora 2 to generate short videos from text prompts.
  The demo highlights easy video generation for quick content and internal communications.
• Governance and management: Microsoft introduced Agent Factory and Agent 365 as control tools for visibility and policy enforcement.
  They help track agent usage, reduce shadow IT, and provide unified billing and support for scaling agents.
• Platform and model updates: Azure AI Foundry now supports third-party models like Anthropic models and links to Copilot Studio without moving tenant data.
  Expect tighter integrations with Work IQ, Fabric IQ, and Foundry IQ to give agents business context and better decisions.

Overview: What Scott Brant Demonstrated at Ignite

At Microsoft Ignite 2025 Scott Brant released a hands-on video that walks viewers through the newest AI features across Microsoft 365. In the footage he demos the new Copilot Agents and the freshly announced Sora 2 integration in Copilot Create, while also showing how the office apps use a new Agent Mode. Consequently, the video serves as a practical guide for anyone who wants to see these tools in action rather than read only the headline announcements.

Brant frames his walkthrough around real tasks: analyzing spreadsheets, drafting documents, automating slides, building basic apps, and generating short videos using natural language. Therefore the content helps viewers understand daily use cases and the limits you might encounter. Also, he mentions licensing and governance updates that organizations need to consider before deploying agents at scale.

Agent Mode in Excel, Word and PowerPoint

First, Brant shows how Agent Mode works inside Excel to analyze data and build formulas automatically, which can save time for routine tasks. However, he also notes that agents require accurate context to avoid incorrect calculations, so human review remains crucial. For Word, he demonstrates drafting and editing assistance that speeds writing, yet he cautions that legal or sensitive text still needs careful oversight.

Meanwhile, in PowerPoint the agent automates slide layouts and suggests content based on a brief, which helps non-designers produce cleaner decks quickly. Nevertheless, Brant points out a tradeoff between speed and creative control: templates and auto-layouts are efficient but sometimes flatten distinct visual identities. As a result, teams should balance automation with manual adjustments when brand or nuance matters.

App Builder Agent, Copilot Studio and Governance Tools

Next, Brant highlights the new App Builder Agent and the enhancements in Copilot Studio that connect to the broader agent ecosystem, including Azure AI Foundry. He explains that these tools let both developers and business users create workflows and small apps, which broadens who can build solutions inside an organization. Yet he stresses that expanding builders brings governance risks, so Microsoft introduced management tools to track and control agent deployments.

In this context, Brant emphasizes Agent Factory and Agent 365 as control-plane offerings that provide visibility, billing, and policy enforcement across agents. Therefore, companies can scale experimentation without losing oversight, but they must invest time to define policies and training to avoid shadow IT. Ultimately, the tradeoff is clear: faster innovation versus the need for stronger governance and monitoring.

Sora 2 and AI Video Creation in Copilot Create

Brant offers a first look at Sora 2 powering AI video generation inside Copilot Create, showing how short clips and visual assets can be generated from natural language prompts. This capability promises to speed up content creation for internal comms and simple marketing materials, while also lowering the barrier for teams without dedicated video editors. However, he warns that AI-generated video currently suits short, generic pieces better than high-quality branded productions, so human refinement will often be needed.

Moreover, Brant demonstrates how Sora 2 can be combined with existing assets to create custom outputs with less manual editing, which can reduce project timelines. Still, tradeoffs include the need for clear rights management and brand checks to avoid misuse of media or inaccurate portrayals. Therefore, while Sora 2 offers potential for productivity gains, it demands governance and careful content review.

Licensing, Small Business Impact and Practical Tradeoffs

Brant also explains Microsoft’s new Copilot licensing model that aims to make access more practical for small and medium businesses. Consequently, smaller teams can pilot Copilot features without the enterprise-level procurement barriers of the past, yet the costs and license complexity still require planning. In addition, organizations must weigh the benefits of productivity gains against recurring licensing fees and integration work.

He further stresses that implementing agents involves operational change: you must map processes, train users, and set monitoring to detect drift or misuse. Therefore the primary challenge is not only technical but also organizational readiness, and leaders should budget time for governance, training, and change management. Ultimately, the best outcomes come from combining clear policies with pilot projects that show measurable value.

Takeaways and Future Challenges

In conclusion, Scott Brant’s video from Ignite 2025 offers a grounded, practical view of how agents and Copilot features can speed routine work and enable new workflows. Yet, he repeatedly cautions that governance, monitoring, and human review remain essential to avoid costly errors and unmanaged deployments. As a result, teams should adopt a phased approach: pilot features, set governance guardrails, and expand only after demonstrating safe, repeatable value.

Looking ahead, Brant believes agents will become more central to how teams work in 2026 and beyond, especially as tools like WorkIQ, Foundry IQ, and cross-platform integrations mature. Nevertheless, the balance between agility and control will define success, and organizations that invest in both training and oversight will gain the most. For readers seeking a practical view of Ignite’s updates, the video offers demonstrations, cautions, and clear examples of tradeoffs to guide real-world adoption.

Keywords

Microsoft Ignite 2025, Copilot Agents hands-on, Sora 2 features, Copilot updates 2025, Microsoft Copilot demo, Copilot Agents tutorial, Sora 2 AI assistant, Enterprise AI tools Microsoft