Microsoft Copilot Super App Relaunch: Home, Code and Autopilot

On September 25, 2026, Microsoft announced a redesigned Copilot app built around three new areas: Home, Code, and Autopilot ^[1]. Executives positioned the redesign as a new operating system for AI-era work, drawing a direct analogy to Office's role in defining work for the PC era ^[2]. That framing is not incidental marketing language - it sets up the strategic logic that makes the relaunch more than a UI refresh. Rather than compete head-on as a fourth or fifth standalone AI chatbot, Microsoft is embedding chat, coding, and autonomous-agent capability directly into a platform that already sits on hundreds of millions of desktops - a battle it can only win by leveraging distribution it already owns, not by out-modeling OpenAI or Anthropic. That is also why the underlying models are treated as interchangeable parts rather than a moat: Cowork runs on Anthropic's Opus 4.8 and Sonnet 4.6, and Frontier-tier customers get access to OpenAI's GPT-5.5, all inside the same Copilot shell ^[3]. The Code tab extends the same logic to software creation itself - it lets any employee describe an app, tracker, dashboard, or automation in plain language and have Copilot build and host it as a shareable internal tool, using the same underlying technology that powers GitHub Copilot for professional developers ^[4]. Nadella has framed the model-agnostic design as insurance against depending on any single AI lab, saying real independence means having a private evaluation suite strong enough to swap in a different model whenever a better or cheaper one appears ^[5]. Put together, the pitch is that Copilot's actual product is not any one model - it is the orchestration layer, the enterprise data connections, and the identity and governance plumbing wrapped around whichever model is winning that quarter.

Autopilot is not a new product so much as a new name with a much bigger mandate. Microsoft first showed an 'always-on personal agent' called Scout, built on open-source OpenClaw technology, at its Build developer conference in June 2026; the rebrand to Autopilot carries that same lineage into private preview at the end of September 2026 ^[6]. What changed between Scout and Autopilot is scope: Microsoft now describes it as a 'digital teammate' with its own identity, memory, computer, and workspace, running on what the company calls Microsoft IQ - an organizational-context layer meant to let the agent reason about how a specific company actually operates rather than answer from generic training data ^[6]. Independent reporting corroborates that Scout's private-preview footprint is what Autopilot is now expanding from, alongside the broader Home and Code rollout arriving through Microsoft's Frontier early-access program in the coming weeks ^[7][8]. The tradeoff Microsoft is explicit about is that a persistent, proactive agent is exactly as useful as it is hard to govern. EVP Jacob Andreou has drawn a specific architectural line against competitors whose agents run inside a local virtual machine on a user's own device: for enterprise customers handling sensitive files, sending that data down to a local VM was 'a complete nonstarter,' so Autopilot's agents are tenant-hosted on Microsoft's side instead ^[5]. That is simultaneously a security argument and an implicit shot at Anthropic's own Claude Cowork product, which uses the local-VM approach Microsoft is calling less enterprise-ready - notable given that Microsoft's own Cowork feature is powered by Anthropic's models ^[3]. Microsoft's own executives concede the underlying tension: agent autonomy and flexibility, the features that make Autopilot valuable, are also what IT admins find hardest to trust, which is why the rollout comes bundled with Agent 365 - a new identity, permissions, and audit layer purpose-built to make autonomous agents inspectable and governable rather than a black box ^[5][6].

Historically, Microsoft sold Copilot the way it sells Word or Excel: a flat per-seat subscription, one price whether an employee used it once a day or fifty times. That model assumed a human being was the bottleneck - a person can only ask so many questions or write so many emails in a workday, so a flat monthly fee against bounded usage made sense. Persistent, proactive agents break that assumption. Cowork, Code, and Autopilot can run continuously, calling tools and consuming compute without waiting for a human to type the next prompt, so Microsoft has layered usage-based billing on top of the per-seat subscription for exactly those agentic features ^[9]. The mechanics: a Copilot Credit costs $0.01 under pay-as-you-go pricing, and how many credits a task burns depends on which model handled it, how much organizational context it had to retrieve, how many tools it called, and how long it ran ^[9]. That structure was already tested - Copilot Cowork reached general availability in June 2026 on this exact metered, $0.01-per-credit model before Microsoft extended it across Code and Autopilot ^[6]. One analysis put the shift in blunt terms: 'the seat has always been a proxy for human throughput,' and once an agent, not a person, becomes the rate-limiting factor, the billing line necessarily has to move with it ^[9]. The catch for enterprise buyers is that Copilot's 'Auto' mode routes tasks to whichever model looks best on accuracy, speed, and cost without publishing the underlying rates, meaning the system, not the customer, effectively decides how fast the meter runs ^[9]. Microsoft's rationale for pushing this now looks straightforward on the numbers: Microsoft 365 Copilot had already crossed 30 million paid seats by July 2026 - still only about 7% of Microsoft's 450-million-plus commercial Microsoft 365 base - with net new seat additions doubling quarter over quarter ^[5][10], and Microsoft 365 commercial cloud revenue hit $100.3 billion in fiscal 2026, up roughly 19% ^[5]. Investors read the relaunch as confirmation of that momentum: Microsoft's stock rose roughly 3.7 to 4% around the announcement, helped by a Stifel upgrade from Hold to Buy issued just two days before the event ^[5][11].

There are two versions of this relaunch happening at once, and they barely overlap. The version in the press release, the Seattle keynote, and the analyst notes is a triumphant platform story: Copilot as the operating system for AI-era work, unifying a fragmented product line - consumer Copilot, Microsoft 365 Copilot, GitHub Copilot-derived tooling, and Scout - that had previously left customers confused about which product did what ^[12]. Coverage even frames the move as Microsoft formally stepping back from chasing the standalone personal-chatbot market head-to-head, choosing instead to concentrate its most capable technology on business customers ^[13]. Corporate VP Annie Pearl described the ambition plainly: Copilot should become 'an all-in-one productivity suite' spanning simple Q&A to fully delegated multistep agent work ^[10]. The other version of this story is unfolding in real time among people who already had the old Copilot app open on their phones and laptops. Reaction concentrated in Copilot's own community forum has run distinctly negative in the days after rollout - not objections to the 'super app' branding or strategy, which most ordinary users never see, but frustration with the lived experience of migration: lost saved conversations and project history with no advance warning, a product that reads as more corporate and less personal than the assistant they were used to, and usage caps appearing almost immediately even for paying subscribers, a tangible surface of the same usage-based billing shift analysts are calling strategically necessary. That gap is not just a rollout hiccup; it echoes the sharpest outside criticism of the launch. Salesforce CEO Marc Benioff dismissed the entire relaunch as a defensive, reactive move, arguing bluntly that 'Copilot's a flop because Microsoft lacks the data, metadata, and enterprise security models to create real corporate intelligence' ^[14]. Whether Benioff's critique is a rival's dismissal or a preview of the adoption friction Microsoft will face at scale, the disconnect between the 'OS for work' narrative and the confused, downgraded experience described by existing users in the first days after launch is the most consequential open question the relaunch has not yet answered.