We were front and center at this year’s Microsoft Build, albeit remotely. Our CEO, MVP Pat McGown, hosted a Build Watch Party and Unconference at the Microsoft Innovation Hub here in Arlington this week, bringing the community together to watch the announcements and talk through what they mean for us as practitioners.
Build is a developer conference, and that means a lot of what was announced was deeply technical developer territory. But if you’re making Microsoft 365 technology decisions on behalf of your organization, this is your heads-up on what’s coming, and what it means for you and your team.
The overarching theme was agent-first. No great surprise there. But the shift underneath it is the part we all need to pay attention to. We’re moving away from using AI agents in a chat window, through delegating bigger tasks to them with Cowork, and now toward agents that do things for us autonomously.
From Copilot to Autopilot
When Copilot first launched, the line everyone repeated was “it’s Copilot, not Autopilot.” AI sat beside you, you stayed in control, and it responded when asked.
At Build, that line disappeared when Microsoft introduced Autopilots: a new category of always-on agents that work autonomously, with their own identity, acting on your behalf. They run in the background and take action without being prompted each time.
This is a real change in posture. We’ve spent the last year getting comfortable with delegating tasks to AI in a chat window, and more recently handing off bigger jobs through Cowork. Now we’ll be able to set up Autopilots that keep working after we’ve closed the laptop.
What Autopilots can actually do
So what can an Autopilot do that Copilot can’t? It comes down to memory and autonomy. An Autopilot is always on. It has its own memory and works inside the Microsoft ecosystem through the Work IQ layer, which gives it the context of how you actually work: who you collaborate with, what you’re working on, and what matters to you.
It works on your behalf behind the scenes: on a basic level monitoring your calendar and rescheduling as conflicts come up, or watching your inbox and Teams messages and responding for you. Of course, they can handle far more than that; you just have to ask.
Sounds great for our colleagues. But red flags for us IT folks. What about security?
An Autopilot acts as its own identity inside your tenant. That puts identity governance, conditional access, and compliance controls front and center. These agents are powerful precisely because they can act, which means the guardrails YOU put around them will matter as much as their capability.
Meet Microsoft Scout: your always-on personal agent for work
Microsoft announced its first ‘Autopilot’ out of the gate, Microsoft Scout, and it’s treating it as a big deal. The pitch is an always-on assistant that surfaces what you need to know and takes action inside your organization’s controls, without you having to ask each time.
Scout works where you work. It joins group chats in Teams, handles threads in Outlook, and reaches across OneDrive and your wider Work IQ context. Think of it as part of a digital team of Autopilots sitting right inside Copilot. Scout will eventually come out of the box, but you’ll be able to custom build more like it. Microsoft is positioning it as “the future of the Copilot ecosystem”.
Right now, Scout is an early experimental release available to Frontier organizations. So, getting your hands on it isn’t a one-click affair. It needs Frontier enrollment, Intune policy configuration, an opt-in attestation, and a GitHub Copilot license. It’s built on the OpenClaw agent framework, with Entra identity governance and Purview compliance layered on top.
This is preview-stage technology for organizations that want to experiment safely, not something landing in every tenant next week. So, if your business is curious, the sensible move is to understand the access requirements and the governance model before anyone switches it on. That’s something we can help you with.
Work IQ APIs: opening up the context layer
The next announcement was about Work IQ and the Work IQ APIs. Work IQ itself isn’t new. It’s the context layer that sits under your Microsoft tools and understands the full picture of your work: your data, your documents, who you collaborate with, and the patterns in how you operate.
What’s new is that developers can now build agents that tap directly into that context layer, including agents built outside Microsoft 365 Copilot. In simple terms, that means custom and third-party agents will be able to reach the same rich organizational context that Copilot starts with today.
For us, that’s a double-edged opportunity. More flexibility to build agents that fit our business, and more reason to be deliberate about data access and of course, governance. The context layer is valuable, which is exactly why you’ll want clear rules about what agents can and can’t read from it.
Web IQ: bringing the live web in
Rounding out what Microsoft is coining “IQ family” is Web IQ. Where Work IQ gives agents the context of your organization, Web IQ gives them better access to the live web.
Agents will be able to pull fresh, verifiable web content alongside your internal context, so their answers reflect what’s true now, not just what’s sitting in your tenant. For anyone relying on agents for research or decisions, that mix of internal knowledge and current web data will be what makes the output trustworthy.
Frontier tuning: teaching AI how your business works
This is something a lot of people have wanted since the beginning. When Copilot first launched, a key selling point was that it didn’t learn from your data. Reassuring for security, frustrating for everyone who wanted AI that actually understood their business.
Satya Nadella framed this one around tacit knowledge: the hard-won, compounding know-how that makes your organization unique. The question he posed was how to preserve and build on that in an age where models can learn almost anything. His answer was to let organizations build their own “hill-climbing machine,” tuned on what they know.
So, soon you’ll be able to tune AI securely inside your own tenant and compliance boundary, teaching it how your organization works, what your processes are, and what it needs to know about your data. Companies including Microsoft, Pearson, and EY have already been piloting it.
The result will be AI tailored to your business rather than generic. The IT consideration, of course, is, again, governance: deciding what it’s tuned on, and keeping that inside your compliance boundary.
Project Solara: agent-first devices
Microsoft gave us a glimpse of where agent-first thinking goes next: the device itself. Project Solara is early-stage (the “Project” label is Microsoft’s signal for experimental work), so nothing is hitting your procurement list soon.
Nadella tied this one back to Microsoft’s original mission of a computer on every desk and in every home, hinting at an updated version for the agent era: agent-first, chip-to-cloud devices.
The idea is a new category of agent-first devices. One example is a small desktop device that proactively surfaces insights, like an upcoming sales opportunity, without being asked. Another is an AI-enabled badge for frontline workers, turning the humble access badge into something with real capability behind it.
Keep it on your radar, not your roadmap. It’s a useful signal of where things are heading.
A word on cost (and the new dashboard)
Now to the question every finance leader will be asking you: what does all this cost?
Microsoft didn’t announce pricing for any of it. Read the room, though, and none of this is cheap to run. As these capabilities arrive, businesses will be paying for them in some form, and the organizations that invest are likely to pull ahead of those that don’t.
There is one practical development on the management side. A cost management dashboard is coming to the Microsoft 365 Admin Center later this year. The Work IQ API will be the first thing tracked there, with more to follow, including cost management for tools like Copilot Studio. That signals Microsoft knows these tools need budgeting and oversight, and that you’ll have a place to keep an eye on the spend.
So, what’s next for you?
While the move from assistant to autonomous agent is underway, it brings with it real questions about identity, governance, and cost along with the capability.
You don’t need to act on any of it tomorrow. But the teams that come out ahead will be the ones who understand what’s coming before it arrives, and who have their governance and identity controls in good shape when it does.
If you’d like to talk through what a well-governed agent-first roadmap looks like for your business, and how to get the foundations right first, that’s a conversation we’re always happy to have.