RSMUS.comMicrosoft Build 2025 felt less like a conference and more like opening the door to tomorrow’s workplace, one where AI agents aren’t just tools, but teammates. As the Frontier Firm report puts it, intelligence on tap “will rewire business,” and Build 2025 showed exactly how. You might be tempted to see these announcements as another round of promising features, but the real story is about moving from “set it and forget it” to a world of continuous, agent-driven collaboration. Here’s how that shows up in three major areas and how you can get ahead of the curve today.
Microsoft isn’t tinkering at the edges; they are laying the foundation for an agentic future. Think of it as evolving from solo performers to an entire orchestra of AI assistants that are tuned, able to swap roles on the fly, and plug directly into your desktop. Here are some of the ways we can leverage to build the future with our clients:
- Copilot Tuning + Multi-Agent Orchestration: AI that learns your playbook and works in teams.
- Azure AI Foundry Local & Dynamic Model Switching: AI you can host anywhere—and swap like building blocks.
- Windows Desktop AI Updates: Agents that see your screen, click your apps, and let you automate the mundane.
These features aren’t about spinning the hype cycle one more time, it’s about giving you a reliable scaffold to build smarter solutions, faster.
- Copilot Tuning + Multi-Agent Orchestration
Big Reveal
Copilot Studio’s headliner at Build 2025 was a one-two punch: Copilot Tuning plus multi-agent orchestration.
First, you can fine-tune Microsoft 365 Copilot on your own data, CRM entries, and playbooks through a guided, low-code interface. On top of that, you can orchestrate and manage multiple agents with each tailored for a distinct phase of your process with human supervision end to end.
Built for Business
With Copilot Tuning (preview), organizations can dramatically shorten their time-to-value, allowing associates to leverage AI that understands their unique business context almost immediately. This means faster project turnaround and quicker ROI for clients.
Meanwhile, multi-agent orchestration amplifies the productivity of each team member, effectively scaling their impact across the organization.
By offloading repetitive and complex tasks to AI agents, firms can scale operations without proportionately increasing headcount, transforming each associate into a high-value contributor capable of delivering more with less effort.
Build Your Toolkit
- Review and Experiment: Review Microsoft Learns and Mechanics video for Copilot Tuning.
- Chain Your Agents: Review Microsoft Multi-agent orchestration and video. Work through the mental exercise of what client scenarios would require you to configure two or three subject matter expert agents to accomplish a goal.
- Azure AI Foundry Local & Dynamic Model Switching
Big Reveal
Microsoft unveiled Foundry Local, a high-performance on-device runtime that brings Azure AI Foundry to Windows and macOS—powered by ONNX Runtime, optimized for CPU, GPU, and NPU hardware for Developers.
Model Router capabilities now let you dynamically switch among multiple cloud-hosted models and automatically choosing the best fit for each task to balance cost, latency, and quality. This further enhances Microsoft’s Multi-Agent workflows.
Built for Business
With Azure AI Foundry Local, businesses unlock capability expansion, enabling them to tackle previously unsolvable problems with on-device AI which is ideal for low latency and strict data privacy scenarios.
Dynamic model switching delivers cost optimization by allowing companies to select the most efficient model for each task in real time, balancing performance and cost-effectiveness.
Together, these capabilities allow mid-market firms to operate with enterprise-level efficiency, achieving a new level of agility and cost control.
Build Your Toolkit
- Master agent fundamentals: Complete Introduction to Microsoft AI agent solutions to understand when to use on-device versus cloud-hosted models.
- Launch an agent in minutes: Work through Get started with AI agent development on Azure using Azure Cloud Shell—build, deploy, and invoke an agent entirely in your browser, no local installs required.
- Deepen your hybrid AI knowledge: Explore AI Agents for Beginners, a 10-lesson Microsoft Learn course that covers core concepts, planning, and hands-on techniques for building impactful AI agents on Azure.
Additional Announcements
- Windows Desktop AI Updates from “apps on screen” to “agents on Windows”
Big Reveal
Computer Use tool (preview) lets Copilot Studio agents see any desktop or browser UI, identify buttons, fields and tables, and then click, type or copypaste just like a human. No APIs or rewrites required.
Model Context Protocol (MCP) and new App Actions land natively in Windows, giving agents a standard, secure way to invoke local app capabilities; developers register actions once and every compliant agent can reuse them.
Windows Subsystem for Linux goes opensource, inviting the community to shape WSL and accelerate crossplatform AI workflows.
Built for Business
These updates let us automate legacy lineofbusiness apps without ripping and replacing—critical for midmarket clients with longtail Windows software. MCPenabled actions provide a governed bridge between desktop workflows and our new agent architectures, while opensource WSL and AIassisted dev tools shrink prototype cycles, letting RSM teams deliver proofsofvalue in days, not weeks.
Build Your Toolkit
- Handson with Computer Use (preview) - Review the Responses API & computerusepreview documentation
- Expose desktop App Actions for PWA - Review Microsoft learns Enable App Actions on Windows
Forging Frontier Firms
Build 2025 didn’t just roll out new features—it handed us the tools to usher our clients into an agentic future, the hallmark of a true Frontier Firm where AI agents underpin every workflow and decision. Now that it’s our turn to step up.
Just as you’ve done before transforming processes, upskilling teams, and delivering breakthroughs that clients once thought out of reach, now is the moment to lead them forward again.
Engage with these announcements and share your learnings, document the value, and help clients realize rapid time-to-value, scaled impact, and expanded capabilities.
The frontier is wide open. Let’s harness these agentic innovations and build the next era of client success.
Seth Bacon
As one of RSM’s Consumer Products industry senior analysts, Seth's job is to understand, forecast and communicate economic, business and technology trends shaping the businesses we serve. Seth has more than 10 years of experience serving clients across the country by providing industry- specific insights and thought leadership focused on digital transformation. Seth works with a variety of clients in the consumer products industry helping them focus on acquiring and retaining customers through data driven insight. Seth's primary focus is on architecting and delivering technology that is intuitive using relevant, real-time information so that individuals can make informed and data driven decisions. Seth consistently works with clients to deliver processes and technology improvements that assist in their technology journey. Seth also helps companies with user adoption through organizational change management and through hands-on training.
