AI product launch: inside Microsoft's Copilot super app

On 25 September 2026 Microsoft unveiled the new Copilot "super app": an AI product launch that gathers chat, code and agents into three tabs, with an explicit comparison to Office.

The move shifts the battleground. The contest leaves the individual tool behind and moves to the single surface where work actually happens every day.

The 25 September launch, in plain terms

Microsoft had teased the app the previous month and made it official on 25 September 2026, as documented in The Verge's account^[1].

The interface brings together three tabs: Home, Code and Autopilot. Home becomes the default entry experience and fuses Copilot Chat and Cowork into a single place.

Inside Home sits Today, a personal dashboard with the emails that matter, meeting invitations and Teams threads. Jared Spataro, Microsoft's chief marketing officer of AI at work, describes it as the point where users orient themselves, review recent activity and pick work back up where they left off.

The Code tab is the surprise, because it lands in an app designed for people who work with documents and meetings. It lets anyone build an app, a tracker, a dashboard or an automation, and share the result with colleagues as an internal, cloud-hosted application.

What this super app really is

The language of the announcement talks about unified capabilities. What the launch is actually about is distribution.

The underlying technology stays familiar: Code runs on the same foundation as GitHub Copilot, inside an isolated environment. The added value lies in the entry point, the default setting and the internal circulation of results.

That difference matters to buyers. A model can be swapped with a different API call; a daily work surface is swapped through a project that runs for quarters and drags in training, permissions and processes.

Microsoft is packaging three separate budgets (conversational assistant, developer tools, agents) into a single line item. The vendor that owns the entry point collects more durable revenue than the vendor with the highest benchmark score.

The Office comparison states the intent

Microsoft believes the new app will carry as much weight as Office. The claim deserves to be read literally.

Office won on bundling and defaults more than on the superiority of any single program. Whoever opened the box already found a spreadsheet, a word processor and email inside, with shared formats and permissions.

The same logic returns here. Chat, code and agents share identity, permissions, history and company context, and the cost of leaving that perimeter grows with every week of use.

This is the clearest signal so far: the game is being played on distribution, the ground where Redmond's historic advantage remains intact. The market has moved from a race on the model to a race on the work surface.

From Scout to Autopilot: the name states the function

In the same announcement Microsoft renamed Scout, the personal assistant shown at Build this year, as Autopilot.

A name change looks cosmetic. Here it declares a different function: Scout evoked research and support, Autopilot promises autonomous execution.

For buyers, the word shifts the question from cost per seat to the perimeter of autonomy granted to the software. An agent that executes requires approval rules, traceability of actions and explicit limits on the systems it can touch.

That issue arrives before the contract does, so it belongs in renewal negotiations. Anyone signing a multi-year deal today is fixing terms on a product set to change function within the life of the agreement itself.

The competitive shift: from feature to surface

GeekWire's^[2] reading frames the operation as an adoption push and as a direct answer to Anthropic and OpenAI.

The axis of competition moves from model quality to ownership of the window where employees spend their day. On that axis Microsoft starts with corporate identity, mail, calendar, Teams and permissions already configured across millions of seats.

Rivals answer with quality, openness and independence. An external assistant stays attractive for companies wary of concentrating on a single vendor, and for those keeping data outside the Microsoft ecosystem.

This publication has long argued a simple thesis: in the enterprise market the defensible advantage lies in deployment rather than in the model. The 25 September launch confirms it, because it puts distribution and integration in the window and leaves benchmarks in the background.

Who feels it in the vendor map

Three vendor categories are already under pressure this quarter.

• Conversational assistants sold per seat, now competing with a tab included in the bundle.
• Low-code platforms for internal apps, overtaken by an app builder inside the productivity app.
• Vertical agent startups, called on to explain their residual value alongside Autopilot.

Price pressure arrives indirectly. A head of procurement who sees three functions inside a licence already paid for will demand steep discounts from anyone selling the same function separately.

Consolidation rewards whoever holds the primary contract. The specialist vendor keeps room when it brings vertical depth, proprietary data or a level of compliance the generalist struggles to guarantee.

There is a concrete counterweight. European companies with data residency constraints and regulated sectors buy by jurisdiction, and on that criterion the vendor map is read by flag rather than by technical ranking.

The counter-argument: announcement and production remain different things

An announcement describes tabs and promises; real adoption is measured in activated licences, weekly users and use cases running in production.

The precedent serves as a warning. Microsoft showed Scout at Build and is renaming it now, a sign the product is still in motion.

The second objection concerns governance. A tab that lets anyone generate internal cloud-hosted apps creates a body of user-built software, with all the audit, permissions and lifecycle burden that follows.

The third objection concerns buying habits. Strength in the consumer market translates poorly into enterprise strength, because the transition demands traceability, data residency and integrations with legacy systems. The enterprise market therefore stays open longer than the noise around the launch suggests.

What to decide in the next 90 days

Many enterprise renewals fall at the end of the fiscal year, so the useful window opens now.

• Chief Strategy Officer: assess which partnership or acquisition covers the layer Microsoft is now entering.
• CFO: review the spending line for tools the new app duplicates.
• Chief Digital Officer: reopen the question of internal-app vendors and per-seat assistants.
• Technology Investor: check which thesis holds when the entry point becomes the moat.

Three concrete actions are worth more than a long strategy: map the overlap between the new app and the current portfolio, ask overlapping vendors for a price review before renewal, and set approval rules for agents before usage spreads.

Negotiating leverage exists as long as the product stays young and adoption data stays thin. Whoever comes to the table holding their own usage numbers gets better terms than whoever comes with an opinion.

The question for next quarter's board meeting fits on one line: how many licences are we paying for today that cover functions the bundle already includes?