Microsoft Fuses Copilot Into One App, Cuts Four Features Before Its September Super App

One app, four casualties

Microsoft began rolling out a unified Copilot app on 13 August, merging the consumer-facing Copilot app and the enterprise-oriented Microsoft 365 Copilot app under a single "Microsoft Copilot" name, icon and web address. Personal, work and school accounts now sit inside the same app with an account switcher, while data, security and compliance controls tied to a work or school account remain separated from personal use, according to Microsoft's own documentation. The rollout reached web and mobile in mid-August, with Windows and Mac apps following in mid-September.

Four features did not survive the merger. Group Chats, which let multiple users collaborate inside Copilot, along with all associated messages and images, was retired on 18 August with no automated migration path, meaning anything a team wanted to keep had to be copied out manually before the cutoff. Copilot Podcasts, the auto-generated audio summaries of meetings and documents, was retired the same day, alongside Deep Research for general consumer use, though existing reports remain accessible through chat history. Mico, the animated character used in consumer voice interactions, is also being discontinued, though elements may persist in education-specific features. Copilot Labs, the experimental-features channel Microsoft used to trial ideas that never reached the main product, is being folded into the wider consolidation as well.

The adoption numbers behind the decision

The merger is not simply a tidiness exercise. It follows an internal memo from Jacob Andreou, the Microsoft executive vice president who now oversees Copilot across consumer and commercial after a March reorganisation, in which he told his roughly 11,000-person organisation that the product needed to "earn the right to exist" in customers' lives.

According to reporting on that memo, an internal audit found that fewer than 4.5 cent of Microsoft 365's 450 million commercial seats had converted to paid Copilot add-ons and that among those who had, only 20 to 30 per cent used the product on a weekly basis.

Microsoft's own research reportedly found that customers disliked switching between multiple separate Copilot products and struggled to identify clear value from the fragmented experience, which is the direct rationale behind folding everything into one surface.

The consolidation is explicitly described as groundwork rather than the finished product.

Microsoft chief executive Satya Nadella confirmed the next step on the company's 29 July earnings call, describing how chat, coding, the Cowork research tool and autonomous Autopilot agents are converging into a single flagship product.

❝

"Copilot is evolving rapidly from chat to Cowork to Autopilots," Nadella told investors. "This quarter, we are bringing these Copilot experiences together, including code, in one super app spanning both consumer and commercial experiences."

Nadella did not give an exact launch date, but the quarter-end target places Microsoft's consolidated offering directly alongside OpenAI's ChatGPT Work and Anthropic's Claude Cowork, in what has become a genuine super-app race among the three largest AI vendors serving enterprise customers.

The wider pattern in enterprise AI product strategy

Microsoft's admission that a low single-digit percentage of its enterprise seats convert to paid Copilot use is not an isolated data point. Across the industry, AI vendors have spent the past two years optimising for feature breadth, adding new capabilities faster than customers could evaluate or adopt them, on the assumption that surface area alone would drive conversion.

Andreou's memo is effectively a public retraction of that strategy from one of the largest players in enterprise software, replacing breadth with a smaller set of features the company believes it can make genuinely load-bearing inside daily work.

For AEC firms evaluating any AI vendor's roadmap, that shift is worth watching for. A vendor cutting features it once marketed heavily is not necessarily a vendor in trouble. It can equally be a sign the company has enough usage data to know which features actually earn their subscription and which were never going to.

Why nearly every AEC firm has a stake in this

Microsoft 365 is close to a default operating layer for architecture, engineering and construction firms. It spans email, document management, meetings and increasingly project coordination, so the Copilot features being retired were not niche experiments for most readers.

A firm that used Group Chats to coordinate a distributed project team or had staff relying on Deep Research to summarise long specification documents, needs to check what is actually disappearing before the migration window closes. Better to find out now than to discover the gap when someone reaches for a tool that is no longer there.

The practical action is straightforward, and it is worth assigning to someone specific rather than leaving it to individual users. IT or operations leads should audit which of the four retiring features their teams actively use before the Windows and Mac rollout completes in mid-September.

The work includes exporting or archiving anything stored in Group Chats, since there is no automated migration path, and confirming whether current Deep Research outputs remain accessible under the merged app or need to be saved separately. None of this is complicated, but all of it has a hard deadline attached.

What the Super App will mean once it lands

The unified app is the visible layer of a larger reorganisation. Andreou's remit now spans design, product, growth and engineering for Copilot across both consumer and commercial lines, reporting directly to Nadella. The broader AI model work sits separately with Mustafa Suleyman, who moved to a narrower research-focused role during the same March restructuring, with Andreou owning the product experience and Suleyman owning the underlying models.

This split mirrors how Anthropic and OpenAI have organised their own consumer-versus-research functions, and suggests Microsoft is deliberately building product velocity independent of frontier model releases.

For AEC firms, the Super App's arrival by the end of September is worth watching for three specific signals once it lands:

Whether Autopilot agents get scoped for professional workflows, such as document review or scheduling assistance, or remain a general-purpose consumer feature initially.
Whether pricing shifts from per-seat licensing toward a seat-plus-consumption model, which would change how firms budget for AI tools tied to Microsoft 365 renewals.
Whether the merged app improves or complicates admin controls for firms running Copilot across mixed personal and organisational device policies, given how central Microsoft's compliance and data controls are to regulated project work.

Takeaway

Audit Group Chats, Podcasts, Deep Research, and Copilot Labs usage across your firm now. The retirement dates are set, and Group Chats in particular, have no automated migration path.
Read the low conversion figure, under 4.5 per cent of seats paying for Copilot, as validation that most firms have not yet found a compelling reason to pay for it, not as evidence the technology lacks value. The merger is Microsoft's attempt to fix that gap.
Track the September Super App launch specifically for agentic features relevant to project delivery, since the current app merger is explicitly a foundation rather than the finished product.
Expect the pricing structure to change alongside the Super App. Firms with Microsoft 365 renewals due in Q4 should build in flexibility rather than locking into current per-seat Copilot terms.

Microsoft's Copilot reset is the kind of quiet infrastructure shift that reshapes how thousands of AEC firms actually work day to day, long before anyone calls it a headline story. Project Flux tracks these changes as they happen so your team is never caught mid-migration without warning. Subscribe to the newsletter to get the September Super App breakdown as soon as it lands.