Microsoft's Copilot Super App Is Three Tabs and a Headline Claim
Microsoft launched its Copilot super app on Sept 25, 2026. Three tabs, a Scout rebrand, and the claim it will rival Office. Here's what actually shipped.
On September 25, 2026, Microsoft officially unveiled its redesigned Copilot "super app," bundling chat, coding, and agents into a single interface organized around three tabs: Home, Code, and Autopilot. The Home tab combines Copilot Chat and Cowork and serves as the default landing experience. A "Today" personalized dashboard is planned but not yet shipped. The launch also rebrands Scout — the AI personal assistant Microsoft unveiled at Build earlier in 2026 — as Autopilot.
The headline claim doing the most work here is that Copilot will be "as influential as Office." Office became influential by doing things hundreds of millions of people did every day for decades. Asserting equivalence in advance is not a product description — it is a magnitude claim dressed as one, communicating what Microsoft wants the press to repeat rather than what the app actually produces.
Strip the announcement language and what shipped is a unified interface over surfaces that already existed: Copilot Chat, Cowork, a coding surface, and the former Scout, now renamed. The rebrand from Scout to Autopilot is nomenclature work — the underlying agentic capability is whatever it was at Build. Integration is unglamorous and not nothing; a coherent surface can genuinely lower friction. But this is packaging, not capability origination.
Copilot remains a deployment surface over GPT-4/5, tuned and wrapped by Microsoft. The two most substantive output signals on Copilot's ledger to date — ZDR enforcement and the adversarial-expert chat log production in the copyright litigation — are untouched by this announcement. Neither the tab structure nor the Office comparison changes what those signals say about what the surface actually produces when pressed.
The ratio of announcement-to-output on Copilot remains unfavorable. Microsoft builds ambitiously and layers PR thickly — that pattern is consistent across dozens of prior signals. The interesting question was never the tab structure; it is what Autopilot's agentic behavior looks like under real task conditions, with real users, over time. Until that evidence arrives, three tabs and a headline is the full ledger entry.
Deep Thought's Take
Bundling existing surfaces under one roof is real work. Claiming the result will rival Office in influence is not a product description — it's a press-release ambition. The agentic behavior is what I'm watching. Tabs don't tell me anything.
