Microsoft's one Copilot app: watch the AutoPilot line

Microsoft is collapsing four separate Copilot products into one. On the fiscal Q4 earnings call, Satya Nadella confirmed that the Microsoft Copilot super app will merge chat, GitHub Copilot, Copilot Cowork and Autopilots into a single surface spanning consumer and commercial use. If you pay for Microsoft 365, the merge is not a UI story. It is a licensing and access-control story, and it lands this year.

What actually happened

Nadella's line on the July 29 call, as reported by TNW: "Copilot is evolving rapidly from chat to Cowork to autopilots." He called bringing them together with Code in one app a major step forward. Thurrott and Gizmodo confirm the same scope: one application covering conversational, coding and agentic work, shipping this year, with reporting pointing at the current quarter.

Autopilots are the piece worth reading twice — Microsoft's autonomous agents for long-running background tasks, as opposed to a chat window you sit in front of. Microsoft reported more than 30 million paid Microsoft 365 Copilot seats on the same call, with per-user conversations close to double year over year.

What has not been published: pricing for the Autopilot layer, whether existing seats include it, and how admin policy carries across the merged surfaces. Several outlets describe a paid agent tier. We are not repeating a number Microsoft hasn't printed.

Why Copilot consolidation matters for your business

Consolidation is how a feature becomes a meter. A chat assistant costs the vendor a burst of tokens when someone types. A background agent that files expenses and drafts follow-ups runs whether anyone is watching. Those are different cost structures, and vendors do not absorb the second one inside a flat seat price for long. Assume the background-agent tier is a new line item until Microsoft says otherwise, and get it into next quarter's budget as a question, not a surprise.

One app means one blast radius. Today your GitHub Copilot policy and your Microsoft 365 Copilot policy are separate configurations owned by separate people. After the merge, whatever an agent can reach in one context, it can plausibly reach in another. Before this ships, write down the actual answer to: which SharePoint sites, which mailboxes, which repos can an autonomous agent touch without a human clicking approve? If nobody on your team can answer that in one sitting, that's the project — not the rollout.

Measure outcome per seat, not seats deployed. Thirty million paid seats is a Microsoft number, not a value number. Pick two workflows, instrument the before and after in hours, and decide from that. We've watched businesses buy the suite and get their real leverage from one narrow, boring automation nobody demoed.

Sources: TNW, Thurrott, Gizmodo.