Your Copilot Just Learned to Work Like Claude — For a Price
Every week now, at least one client asks me some version of the same question. "We've got Copilot. We've got Claude. Do we actually need both?" They're not really asking about the tools. They're asking whether they're paying twice for the same thing.
Up until a few weeks ago, my honest answer was: not exactly, but close enough to sting. Copilot could draft and summarize. Claude, through its own Cowork product, could go do things — plan a task, work across files and messages, hand back something finished instead of a first pass someone still had to finish. Two tools, two different levels of ambition, and every business I work with that runs both was paying for the gap between them.
Microsoft Just Closed the Gap — With a Familiar Engine
Copilot Cowork reached general availability worldwide on June 16, 2026, after three months in Microsoft's Frontier preview. The pitch is straightforward: describe an outcome, and Cowork plans and executes the multi-step work — across your files, your email, your chats — and hands you back a finished deliverable instead of a draft. It's the same agentic move Claude's own Cowork already makes. Microsoft built a version that lives inside Copilot.
Here's the part I make sure every client hears, because it never makes the announcement headline: at general availability, Cowork runs on Anthropic's Claude models — Opus 4.8 and Sonnet 4.6. If you've been running Claude and Copilot side by side specifically because Claude was the one that could actually finish a job, Cowork just moved that capability into the tool you're already paying for. I tell clients already invested in both that this is a real consolidation opportunity. I also tell them it's metered, and the way it's billed is worth understanding before an admin turns it on tenant-wide.
What You Actually Get When Cowork Is Turned On
Two capabilities come with it: task scheduling, so Cowork can run work on a recurring basis instead of only on demand, and the ability to build Skills — the same concept Claude uses, scoped to Cowork only, not portable elsewhere in Copilot. Both require an existing Microsoft 365 Copilot license. Both are off by default. An admin has to turn Cowork on before anyone in the tenant can touch it, which is exactly where I tell clients the rollout decision belongs — with IT, not with whoever asks for it first.
Where the Bill Actually Comes From
A $30-a-user-per-month Copilot license gets you in the door. Every Cowork task after that draws from a separate, billed credit pool. Microsoft's own documentation shows simple tasks running $0.70 to $2, and complex, multi-step tasks at $15 or more. I ask clients to model ten complex tasks a week before they commit to anything — that's roughly the point where this stops being a rounding error and starts being a line item someone has to explain to ownership.
The Cutoff Nobody Warned Admins About
Billing enforcement began July 1, 2026. Tenants without usage-based billing configured lose access to Cowork entirely — a hard cutoff, not a warning banner. Microsoft auto-provisions the billing infrastructure during admin setup, and admins can set spending caps once it's configured, so there's real ability to control exposure. But it has to be set up deliberately, and it's the first setting I walk through with a client before I let them turn Cowork on for the whole tenant.
If you've read the post about how much of Microsoft 365 goes unused before anyone finds it, you already know the shape of this problem. The capability ships. The configuration that makes it usable — and in this case, affordable — is the part that requires someone to actually go set it up.
Who I Think This Actually Matters For
Two groups. First, businesses already paying for both Claude and Microsoft 365 Copilot who want to consolidate some of that work into one interface — Cowork is the first real bridge between the two, and it's the group I expect to move fastest. Second, any Copilot-licensed business trying to decide whether Cowork's per-task cost is worth it for how their team actually works, before an admin flips it on for the whole tenant. That second decision deserves more thought than most feature toggles get, precisely because this one has a meter running behind it. Worth reading what Copilot itself actually does before Cowork enters the picture if you haven't already — Cowork sits on top of that foundation, it doesn't replace it.
What I Don't Know Yet
Most of the public case studies so far skew enterprise — Fortune 500 adoption, enterprise plugin partners, the kind of usage volume that makes a $15 task disappear into a much bigger budget. The real-world cost-to-benefit picture at smaller scale is still thin, and I'll say so plainly rather than pretend otherwise. That's not a reason to skip it. It's a reason to run it deliberately, with spending caps set before day one, instead of letting the first invoice tell a client what they should have configured.
Frequently Asked Questions
Do we need to buy anything new to get Copilot Cowork? No new license. If you already have Microsoft 365 Copilot, Cowork is a capability that sits on top of it. It's off by default, and an admin has to turn it on and configure usage-based billing before anyone in the tenant can use it.
What happens if we don't set up billing? Nothing gets billed by accident — but nothing works either. Billing enforcement began July 1, 2026, and tenants without usage-based billing configured lose access to Cowork entirely. There's no grace period. It simply doesn't run until billing is configured.
How much should we actually budget for this? It depends entirely on how the team uses it. Microsoft's documentation puts simple tasks at $0.70 to $2, and complex, multi-step work at $15 or more. I tell clients to model ten complex tasks a week before rollout — that's roughly where it turns into a real monthly line item, which is exactly why setting a spending cap before rollout matters more here than it does on most Copilot add-ons.
Copilot got more capable. It also got a meter. I'm telling every client to look at both before they flip the switch.
Accurate as of July 2026. Microsoft updates its products and pricing regularly.
J. Scott Clark is the President and CEO of The 365 Collective, Inc., a Microsoft 365 consulting and training firm serving small and mid-sized businesses across healthcare, finance, construction, engineering, publishing, and retail.
This is exactly the kind of decision we walk clients through before they turn a feature on tenant-wide, not after the invoice shows up. If you want a hand figuring out whether Cowork's economics make sense for how your business actually works, feel free to reach out.