On 9 July, OpenAI launched GPT-5.6, the latest model in its Sol/Terra/Luna frontier family, and Microsoft immediately designated it the preferred model across five core Copilot surfaces: Word, Excel, PowerPoint, Copilot Chat, and Cowork. Day zero. No waiting period, no staged rollout announcement, no "coming soon." The model shipped and the enterprise got it simultaneously.

That's a meaningful moment, and it deserves more than a changelog entry.

What actually changed, and why it matters

GPT-5.6 brings measurable improvements in two areas that senior leaders care about most: output quality and agentic task completion.

Output quality means the prose Copilot drafts in Word is more coherent, the summaries in Copilot Chat are more accurate, and the analysis generated in Excel is better reasoned. These aren't dramatic leaps that will shock anyone. They're the kind of steady, compounding improvements that quietly raise the floor of what your teams produce every day.

Agentic task completion is the more interesting story. Copilot Cowork benefits directly when the underlying model gets smarter at multi-step reasoning. Think of it this way: if you asked a capable colleague to reorganize a folder of quarterly reports, tag them by region, and surface the three with the biggest variances, they'd need to hold context across several steps without losing the thread. A better model does that more reliably. That's the practical difference GPT-5.6 makes in Cowork specifically.

For leaders, the headline is simple: the productivity tool you're already paying for just got better, automatically, with no migration, no retraining, and no additional cost.

The partnership signal hiding in plain sight

There's a second story worth paying close attention to, and it's the one Microsoft didn't put in the press release.

Bloomberg reported recently that Microsoft has been quietly routing some Copilot prompts to its own MAI models — internal AI models built to reduce Microsoft's dependence on OpenAI infrastructure and, frankly, to manage costs. The GPT-5.6 announcement landed the same week, and the timing was not accidental. Publicly designating a brand-new OpenAI model as the preferred default across Copilot was, among other things, a clear reaffirmation of the Microsoft–OpenAI partnership at a moment when the market was asking questions about its durability.

Here's the nuance leaders should hold: both things can be true simultaneously. Microsoft can be a committed OpenAI partner for frontier-model capability while also investing in its own model infrastructure for efficiency and resilience. The MAI routing story signals a long-term cost strategy, not a break-up. But it does mean the model powering your Copilot investment may not always be the one on the label — and that's worth monitoring as enterprise AI procurement conversations mature.

For now, GPT-5.6 is what's running. The quality is there. The partnership is intact, publicly at least. But keep one eye on how Microsoft's AI sourcing strategy evolves over the next 12 months.

What this means for leaders

The practical actions here are straightforward, but they're worth making explicit rather than leaving to chance.

• Communicate the upgrade proactively. Your teams may notice improved output in Word, Excel, and Copilot Chat without understanding why. A short internal message that names the improvement and connects it to the investment already made builds confidence and encourages continued adoption. Don't let a genuine improvement go unacknowledged.

• Revisit your Cowork pilots. If you've been evaluating Cowork for task automation and found the agentic reliability underwhelming in earlier testing, this is a good moment to run those scenarios again. A stronger underlying model directly improves multi-step task completion. What didn't work before might work now.

• Track the MAI routing story. Assign someone to monitor how Microsoft's model sourcing evolves. As enterprise AI contracts come up for renewal, understanding what model is actually processing your data matters for quality benchmarking, vendor conversations, and governance documentation.

• Don't wait for "perfect." The upgrade happened. The value is available today. The organizations that extract the most from enterprise AI are the ones that engage with it continuously rather than waiting for a future state that looks more certain. It never does.

The bigger picture

What happened on 9 July is a small but instructive example of how enterprise AI now works. The frontier moves fast. The productization is instant. The competitive and commercial dynamics between major players shape what's under the hood of the tools your teams use every day, whether or not anyone mentions it in the release notes.

The leaders who will navigate this well aren't necessarily the ones with the most technical knowledge. They're the ones who stay curious, ask the right questions of their vendors, and build internal habits of paying attention.

GPT-5.6 is a good upgrade. Take the win, communicate it to your teams, and then keep watching. There's a lot more movement coming.

For the last few years AI has handed us drafts. Cowork hands back finished work and that changes the question from "will people use it?" to "what do we trust it to own?"

Most AI at work still follows one pattern: you ask, it answers. Helpful, but you're still holding the pen.

On 16 June, Microsoft moved that line. Copilot Cowork is now generally available worldwide and it doesn't give you a draft. You define the work, and Cowork runs the whole task end to end and returns a completed result.

What it actually is

Not a chatbot. Not a smarter prompt box. Cowork is an agentic system built for complex, long-running, multi-tool tasks, the kind that used to eat a week.

A few real examples from the preview:

• 📊 An engineer taught it to safely edit batch-job spreadsheets and redraw the dependency charts after every change. Automatically.

• 🗂️ One team compared nearly 4,000 files across two product versions. Weeks of work, done.

• 📉 A sales lead pointed it at a stalled pipeline and got back a ranked list of at-risk deals with the exact follow-up that had gone cold on each. A week collapsed into a morning.

After three months in preview, more than half of the Fortune 500 were already using it.

Why it works

• ☁️ Runs in the cloud — tasks keep going even when your laptop's off.

• 🧠 Grounded in Work IQ — it uses your real business context, not a generic guess.

• 🔒 Stays inside your Microsoft 365 trust boundary — your existing policies and controls still apply.

• 🔀 Multi-model — the right model gets matched to the task; you're not locked into one.

The question every leader will ask: what does it cost?

Here's the shift. Your Copilot license stays a predictable monthly fee. Cowork sits on top and is billed by usage in Copilot Credits, priced per task.

The good news for anyone planning a budget:

• 💷 Pay-as-you-go at $0.01 per credit, or commit volume up front for a discount.

• 📐 Microsoft published a free estimator so you can model your own spend before you commit.

• ⏸️ Cowork is off by default. Admins decide when it turns on and who gets access.

• 📊 Spend limits and alerts at tenant, group, and user level — governance built in, not bolted on.

Bottom line

GA isn't the headline. The shift is.

We're moving from Copilot as assistant to Copilot as a colleague you delegate to. That flips the questions every organization now has to answer: which work do we trust it to own, how do we govern the spend, and how do we measure what it gives back?

The companies that answer those three first won't just adopt a feature. They'll redefine what a productive day looks like.

The shift is here. The advantage goes to whoever operationalizes it on purpose.

So I put together this infographic to cut through the noise. Three tools. Three very different jobs. And one simple way to remember which is which based on who initiates the work: you, or the AI.

Have a look 👇

Yesterday, Microsoft changed that equation.

Meet Microsoft Scout: the first Autopilot agent

Microsoft Scout is not a chatbot. It's not a smarter search box. It's the first example of a new category Microsoft is calling Autopilots. Always-on agents that run in the background and take action on your behalf, without needing a prompt every single time.

Think of it as a highly capable colleague who's always in the loop: monitoring your calendar, watching your inbox, tracking your commitments and quietly handling coordination before it lands on your plate.

Less coordination tax. More focus time.

Scout lives inside the apps you already use: Teams, Outlook, OneDrive, SharePoint. Here's what it actually does:

• 📅 Meeting coordination: Schedules across time zones, flags what matters, and preps your briefing materials automatically.

• ⏱️ Time blocking: Spots upcoming deliverables and reserves focus time on your calendar before someone else takes it.

• ⚠️ Risk surfacing: Detects stalled decisions and flags them early, before they become actual blockers.

• 🧠 Work IQ: Learns how you work over time, becoming more relevant and aligned to your priorities with every interaction.

You stay in control

One thing worth saying clearly: Scout doesn't go rogue. It operates within your organisation's existing security and compliance policies. The same rules that govern everything else in Microsoft 365. Sensitive actions can require human sign-off before anything happens. You stay in the loop, you stay in control.

Why this matters beyond the announcement

The most interesting part of Scout isn't any single feature. It's the shift in mindset it represents. We're moving from AI that answers questions to AI that carries work forward.

For anyone managing productivity and collaboration tools today, this signals where things are heading. The question to ask now isn't "should we pay attention?" It's "how do we get ready?"

💡 Availability: Microsoft Scout is currently in private preview for Frontier organisations. → Full details and setup instructions

A few weeks ago I was chatting with colleagues in Legal about how they take meeting minutes. Their workflow surprised me.

A note-taker captures the discussion by hand. Afterwards, they paste those notes into a Word document and ask Copilot to polish them into a clean summary. Manual capture, AI cleanup at the end.

When I asked why they didn't just use the transcription and recap features built into Teams, the answer was immediate: those meetings can't be recorded or transcribed. Legal and compliance reasons. Full stop.

That stuck with me. Because the meetings that would benefit most from AI help (Legal discussions, HR conversations, M&A, executive sessions, investigations) are precisely the meetings where the technology isn't allowed to listen in the usual way.

So I went looking for whether anyone was solving this. It turns out Microsoft is.

The shift that changes things

Until now, getting an AI summary of a meeting meant accepting a trade-off: you had to let Teams record or transcribe it first. The summary was built from the transcript. No transcript, no summary.

That's about to change.

A new capability is rolling out in Teams that lets Copilot generate a meeting recap without saving a transcript or recording. The AI listens in the moment, produces the summary, and when the meeting ends the underlying conversation simply isn't kept. You walk away with the recap. Nothing else stays behind.

It sounds small. It isn't.

Why this is bigger than it looks

For two years, the conversation inside most companies has been some version of "AI features or compliance policies, pick one." Highly regulated meetings stayed locked out of the productivity wave, because the only way to unlock AI summaries was to retain content that legal teams couldn't allow.

This new approach reframes the question. It says: both. You can have the AI value in the moment without keeping the source material afterwards.

For my Legal colleagues, this means the cognitive load of being a human transcription machine starts to lift. They get to listen and participate again, instead of typing while everyone else talks. The Word-document detour at the end of every meeting becomes optional rather than mandatory.

And for the rest of the business, it opens up AI assistance for entire categories of conversations that were previously off-limits.

A word of caution before you celebrate

A few things worth keeping in mind.

This isn't "AI with no data at all." Even without a transcript, the summary itself is content and like any content in your Microsoft 365 environment, it lives inside your governance and retention policies. Treat it accordingly.

It's also worth being deliberate about which meetings use this. "We can now use AI everywhere" is a different policy than "we should use AI everywhere." Some conversations are sensitive enough that even a summary deserves a deliberate choice rather than an automatic one.

The bigger pattern

What I take away from this isn't really about a single feature. It's that AI tools are slowly learning to bend toward enterprise reality instead of fighting it. The early Copilot era was about capability. The next era is about fit: making the technology work inside the legal, regulatory, and cultural constraints that real organizations live with.

This isn't a marketing tweak. It changes how you should think about who in your organization actually needs the Teams Premium add-on, and which features your existing Teams Enterprise users can start using today at no extra cost.

Here's the executive summary, the practical detail, and what I'd do about it.

Microsoft has split the previous Teams Premium feature set into two buckets:

• Now included with Teams Enterprise → Most large-scale events capabilities (town halls, webinars, immersive events) and the new end-user features for Microsoft Places.

• Still requires Teams Premium → The intelligence, protection, branding, advanced collaboration, advanced Bookings, and the Queues app capabilities.

In short: events go mainstream, while the AI, security, and contact-center muscle stay premium.

What moved to Teams Enterprise (the wins)

The biggest shift is around events at scale. If your communications, HR, or marketing teams have been improvising with vanilla meetings because Premium wasn't budgeted, that workaround is over.

Town halls: production-grade, no add-on required

Teams Enterprise organizers can now run interactive town halls for up to 3,000 attendees with reactions, raise hand, polls, Q&A and view-only events for up to 10,000 attendees. The full polish is included: streaming chat, Microsoft eCDN for bandwidth control, Ultra-Low Latency streaming, 1080p video, town hall insights, and custom-branded emails.

Need to go bigger than that? You'll buy an Attendee Capacity Pack (sold separately, scaling up to 100,000 attendees) but you no longer need a Teams Premium license per organizer just to unlock the feature set.

Webinars: every "advanced" feature is now standard

The list of webinar capabilities that moved is long: custom emails and reminder send times, waitlists for sold-out registrations, registration time windows, manual registrant approval, controlling what attendees see, hiding attendee names, and RTMP-In for producing events from external encoders. Attendance and engagement reports come along for the ride.

If you'd previously written off webinars as "nice but not worth the licensing math," it's worth a fresh look.

Immersive 3D events and Microsoft Places

Two more meaningful additions: immersive events in Teams (3D environments for up to 300 attendees, including Meta Quest support) move into Teams Enterprise, and the end-user features of Microsoft Places (Places Finder for richer room booking and Places Explorer for map-based reservations) become available with any license that includes the Outlook/Teams calendar.

That last one is broader than it sounds. It means Microsoft 365 E3, E5, Business Standard, Business Premium, and several other SKUs all benefit. For organizations rethinking hybrid work, that's a free uplift.

What's still gated behind Teams Premium

Microsoft was explicit on this point: Teams Premium continues to exist, and the features that remain are arguably the most strategically valuable ones.

Intelligent Meeting Recap and AI-powered communication

This is the single biggest reason most organizations bought Teams Premium in the first place and it's not moving. AI-generated notes and tasks, intelligent meeting recap (including for VoIP and PSTN calls), live translated captions and transcripts, speaker timeline markers, autogenerated chapters, and multilingual recap all stay in Premium.

If your organization has built workflows around AI-summarized meetings, the licensing math hasn't changed.

Advanced meeting protection

For regulated industries, this is the non-negotiable bucket: watermarking, end-to-end encryption (up to 200 participants), Microsoft Purview sensitivity labels for meetings, detect sensitive content during screen sharing, and granular control over who can record and transcribe. These remain Premium-only.

The Queues app for Teams Phone

If you're running Teams Phone as a contact-center-lite, the Queues app (call queue and auto-attendant management directly in Teams, real-time metrics, historical reporting, and silent coaching) stays Premium (and still requires a Teams Phone license).

Branding, advanced collaboration analytics, and advanced Bookings

Custom meeting themes, organizational backgrounds, custom Together mode scenes, and bespoke meeting templates remain Premium. So do the advanced collaboration analytics for IT (insights on inactive teams, external collaboration patterns), audio/video quality alerts, custom user policy packages, and the SMS-notification and queue-view enhancements for Bookings and virtual appointments.

So what should you actually do about it? Three concrete moves I'd make this quarter:

1. Re-scope your Teams Premium license footprint. If you bought Premium primarily so a handful of organizers could host large town halls, you may be paying for something that's now free. Map your current Premium assignments to the reasons people have the license and reassign accordingly.

2. Run a "what can we turn on tomorrow" audit. Town hall capabilities, webinar registration controls, and Places Finder are sitting there waiting in your Teams Enterprise tenant. If those features have not been announced yet, this is the perfect timing to do so.

3. Reframe Teams Premium as your "AI + protection + voice" license. That's the value proposition now. If a user needs intelligent recap, watermarked confidential meetings, sensitivity labels, or the Queues app, they need Premium. Otherwise, they probably don't.

Bottom line

This change is a quiet win for most Microsoft 365 customers. The features that drive everyday collaboration at scale are now part of the base experience, while Teams Premium sharpens its identity around AI, security, and voice.

If you're heading into a renewal cycle, this is the moment to do the math.

For most of us, that is a meaningful change.

What is actually new:

→ A dedicated Deep Thinking mode inside Copilot, selectable when a task deserves more rigor

→ Stronger performance on multi-step reasoning: analysis, strategy, technical design, long-form drafting

→ The same Copilot experience you already know for everything else. Speed where you need it, depth where it matters

And here is the shift worth paying attention to:

For the past few years, the AI conversation has been about faster. Faster answers. Faster drafts. Faster summaries.

But the questions that actually shape our work (the strategy memo, the contract review, the architectural decision, the difficult email) were never really about speed. They were about judgment.

Deep Thinking changes the equation in three ways:

→ The bottleneck moves. It is no longer "how fast can AI answer?" but "how well can I frame the question?"

→ Our role evolves. We become editors and challengers of AI reasoning, not just consumers of AI output.

→ The value shifts upstream. The hardest, most ambiguous problems (the ones we used to keep away from AI) are exactly where these models now earn their place.

The teams that will get the most from GPT-5.5 are not the ones using it for more tasks. They are the ones using it for the right tasks and knowing the difference.

A few use cases get identified in workshops, someone builds a pilot, results look promising, and then... nothing scales. The pilot team moves on, the business sponsor loses visibility, and six months later leadership is asking (again) why the AI investment hasn't translated into measurable outcomes.

The problem isn't the technology. It's the absence of a framework that takes a use case all the way from intake to business as usual with clear gates, defined ownership, and executive alignment at every stage.

After contributing to AI adoption in a global enterprise environment, I've refined a three-phase framework that addresses exactly this. It's the same structure I'll be using as the foundation for the next posts in this series.

The Framework at a Glance

Three phases. Three milestone gates. One goal: turning AI capability into sustained business value.

Phase 1 — Intake & Journey Design → M1 Gate: Capture the need, validate feasibility with Microsoft advisory, define the business outcome and AI journey strategy, and secure executive sign-off on the blueprint.

Phase 2 — Enablement & Pilot → M2 Gate: Establish governance, prepare the organization, run a controlled pilot, measure against defined KPIs, and decide whether to scale, adjust, or stop.

Phase 3 — Scale & Run → M3 Gate: Roll out in waves, activate a champion network, embed the capability into BAU support, and operate with continuous measurement and improvement.

Each gate is a formal SteerCo checkpoint. No gate, no next phase. This discipline is what separates AI programs that scale from those that stall.

Why Gates Matter

In my experience, the biggest risk to enterprise AI adoption isn't bad use cases. It's use cases that move forward without proper validation.

Gates force the conversation. At M1, the SteerCo validates that the blueprint is aligned with business priorities before a single pilot resource is spent. At M2, the SteerCo reviews pilot outcomes against pre-defined success criteria before committing to scale. At M3, the SteerCo confirms that the capability is ready to operate under a BAU model with measurable ongoing value.

Without gates, good intentions turn into sunk costs. With gates, every phase has a clear decision point and accountable ownership.

The Role of Microsoft and the SteerCo

Two external reference points appear at specific stages of the framework:

Microsoft Advisory: positioned during Use Case Validation in Phase 1. This is where you bring in Microsoft's technical and adoption expertise to validate that the use case is feasible, the capability fits the Copilot ecosystem, and the approach aligns with Microsoft's best practices. Skipping this step is how organizations end up trying to solve problems that Copilot isn't designed to solve, or missing out on capabilities that would have been a better fit.

SteerCo Checkpoints: appear at three critical moments: Blueprint Definition (end of Phase 1), Pilot Exit Decision (end of Phase 2), and Operate, Measure & Improve (ongoing in Phase 3). The SteerCo isn't a ceremonial review, it's the governance body that ensures executive alignment, unblocks escalations, and makes go/no-go decisions on AI investment.

Why This Matters Now

With the pace at which Microsoft is shipping new AI capabilities (Copilot Cowork, Researcher agents, agent mode across Office apps) organizations are under pressure to move fast.

But speed without structure creates risk. A framework doesn't slow you down; it protects the investment you're making and ensures the features you enable actually deliver value for the business.

If you're building or refining your organization's AI adoption approach, follow along. The shift starts with structure.

The infographic beneath maps the full framework visually. Feel free to share it with your team or use it as a reference when designing your own adoption approach.

Questions or experiences to share? Drop a comment or connect with me on LinkedIn.

I've seen it working live. Multi-step task orchestration across M365 apps, autonomous plan generation, visible progress with user control. The comments on social media are right: it's a game changer (and also a change of mindset).

But here's my concern.

I'm sitting on an EU tenant. And Copilot Cowork requires Anthropic to be enabled as a Microsoft subprocessor. For EU/EFTA and UK tenants, that toggle is off by default. And for good reason:

⚠️ Anthropic AI models are NOT in scope for EU Data Boundary commitments.

⚠️ Data inferencing occurs outside the EU Data Boundary.

⚠️ No customer data is stored outside the EU, but processing happens in the US.

So before you flip that switch, you need to understand what you're accepting.

And this is where I see a pattern that concerns me.

💡 The pace at which Microsoft is releasing major updates reminds me of the crypto bull runs during COVID-19. Every week a new token launched with incredible promises, a strong team behind it, and massive FOMO driving investment decisions. Very few people stopped to DYOR (Do Your Own Research).

The parallel is uncomfortable but real:

→ Hype-driven announcements creating urgency to enable features immediately.

→ FOMO pushing organizations to adopt before assessing the implications.

→ Little room for a structured evaluation of data governance, compliance, and alignment with your existing roadmap.

This isn't a criticism of the technology. Copilot Cowork is genuinely powerful. The multi-model strategy Microsoft is building with OpenAI, Anthropic, and soon likely others, is the right direction for the enterprise.

But enabling a major change without understanding the data governance implications is not an adoption strategy. It's a risk management gap.

🎯 What I'd suggest for organizations navigating this:

1️⃣ Have an adoption plan. Don't enable features because they're new. Enable them because they fit your roadmap and your users are ready.

2️⃣ Stay the course. Continue delivering your existing roadmap. New features are valuable when they're introduced at the right time, not the first time.

3️⃣ Do your own research. Evaluate what enabling Anthropic as a subprocessor means for your tenant. Review your DPIA posture, your data residency commitments, and your sector-specific regulations. For EU tenants, the EU Data Boundary exclusion is not a detail, it's a fundamental consideration.

4️⃣ Don't let hype drive governance decisions. The gap will be filled. Microsoft will likely expand Anthropic's data residency coverage. But until it does, understand exactly what you're enabling and what you're accepting.

The best adoption strategy isn't the fastest one. It's the one that delivers value without compromising trust.

Copilot Cowork will still be there when your organization is ready for it. ⏳

#Microsoft365 #Copilot #CopilotCowork #DataGovernance #AIAdoption #EnterpriseAI #EUDataBoundary #TheAIShift

Three months later, the dashboard tells a familiar story: 20% of licensed users are active, the rest tried it once and moved on. Leadership starts asking uncomfortable questions about ROI.

The problem isn't the technology. It's the gap between deploying a tool and getting people to actually change how they work.

After contributing to the Copilot adoption in a global enterprise environment, I've learned that the missing piece isn't better training or more licenses. It's something most rollout plans completely overlook: an internal adoption brand.

The Pattern I Keep Seeing

Here's how it typically goes. Copilot gets turned on. Users see a new icon in Word or Teams. The curious ones try a prompt, get an underwhelming result because they don't know how to prompt effectively, and conclude that "Copilot doesn't really work." The rest never try at all.

There's no shortage of capability. There's a shortage of connection. A clear, trusted channel that helps users understand why this matters, how to start, and where to go when they get stuck.

That's the gap. And no amount of licensing fixes it.

What an Adoption Brand Actually Does

An adoption brand is not a logo. It's a recognizable identity that wraps everything related to your Copilot rollout into one coherent experience: communications, training, tips, success stories, support channels, and executive messaging.

Think of it this way: when users hear the brand name, they know exactly what it means and where to go. It becomes the single entry point for everything related to Copilot in your organization.

What this looks like in practice:

A name and identity that people recognize across emails, Teams channels, intranet pages, and events. Not "IT Copilot Rollout Phase 2." Something human that sticks.

Regular communications in the users' language. Not technical documentation, but practical tips, quick wins, and real stories from colleagues who found value. Consistency matters more than volume.

A feedback loop where users can share what works, what doesn't, and what they need. This turns passive recipients into active participants and gives you adoption data you won't find in any dashboard.

The brand becomes the bridge between the technology and the people who are supposed to use it.

The Three Stages

In my experience, organizations that build an adoption brand go through three distinct stages:

Stage 1 — The Gap. Copilot is deployed and licenses are assigned, but adoption stalls. Without clear guidance, users disengage early and the investment goes underutilized. Technology is in place. User engagement is not.

Stage 2 — The Bridge. A dedicated adoption brand is launched. A single, recognizable channel for communications, training, and success stories that drives user connection with Copilot. One brand. One channel. Clear user connection.

Stage 3 — The Shift. The adoption brand is established. Users engage consistently, resistance decreases, and Copilot usage scales across the organization with measurable impact. Sustained adoption achieved. Copilot embedded in the daily work.

The progression from gap to shift doesn't happen overnight. But it happens significantly faster when there's a brand holding it all together.

What's Next

This post is the first in a series on enterprise AI adoption. In the coming weeks, I'll share the full adoption framework for targeted scenarios, from intake and journey design through pilot execution to scaling across the organization.

If you're building a Copilot rollout plan or struggling with adoption in your organization, follow along. The shift starts with how you show up to your users.

This infographic maps the three stages visually. Feel free to share it with your team or use it in your internal planning.

Questions or experiences to share? Drop a comment or connect with me on LinkedIn.

If you manage Microsoft 365 in any capacity, you need to know about MC1253858. Starting April 15, 2026, Microsoft is splitting Copilot into two clearly different tiers. And the gap between them is significant.

Here's what's happening, what it means, and what you should do about it.

The Split

Microsoft is introducing two labels that will appear inside the products:

• Copilot Chat (Basic): users without a Microsoft 365 Copilot license

• M365 Copilot (Premium): users with a paid license

The biggest change: Copilot disappears from Word, Excel, PowerPoint, and OneNote for Basic users. The in-app side panel that rewrites paragraphs, generates formulas, and builds slides is gone unless you have a premium license.

Smaller orgs, take note: if your tenant has fewer than 2,000 users, see MC1253863. Copilot stays in Office apps but under throttled "standard access" rather than full removal.

What Basic Users Keep

It's not all gone:

• Microsoft 365 Copilot App: standalone AI chat, file upload, image generation, Copilot Pages, plus Word/Excel/PowerPoint agents for chat-first content creation

• Copilot in Outlook: inbox and calendar grounding stays intact

• Pre-built agents from the Agent Store (Prompt Coach, Writing Coach, etc.)

• Agent Builder Lite: create simple personal agents with public web and SharePoint as knowledge

• Enterprise Data Protection: still applies to all interactions, regardless of tier

What Basic Users Lose

• In-app Copilot in Word, Excel, PowerPoint, OneNote: no more side panel assistance while working in documents

What You Should Do Now

Define the strategy to maximize the value of the premium license. This is no longer "give everyone Copilot and see what happens." Analysts, content creators, project managers, and executive assistants are likely your highest-ROI candidates.

Update your training materials and communicate the change before April 15. If your internal guides reference Copilot in Word or Excel as generally available, they become misleading in the upcoming weeks. Get ahead of the confusion.

Promote the M365 Copilot App. It's the primary AI tool your Basic users will have, and most people is not getting most of value out of it. The chat-first creation workflow with agents is genuinely powerful, but only if people know it's there.

Measure actual usage now. Real usage data should drive your license distribution, not seat counts. Use the next few weeks to identify the super users relying on the in-app experience and build a plan around it.

The Bottom Line

This is Microsoft monetizing AI adoption after a period of broad free access. The in-app experience is genuinely more valuable than the chat-first alternative, and that gap is now explicit and paid.

For organizations with a clear strategy about AI-powered productivity, the premium license will likely pay for itself in the right roles. For everyone else, the Copilot App and Outlook integration still offer real value but the line has been drawn.

April 15 is around the corner. Plan accordingly.

I've created a simple but powerful comparison infographic mapping the main capabilities across both tiers. Grab it at the top of this post and share it with your team if it's of your interest.

Questions or clarifications about it? Drop a comment or reach out on LinkedIn.