Microsoft brings full Office into a merged Copilot app for the first time

"How did a multinational company do a forecast in 1981? You had interoffice memos and faxes and all of these artifacts that then just disappeared and you suddenly had an Excel spreadsheet in an email," Satya Nadella said at Microsoft's customer-only event last week ^[3]. As The Verge describes it, he is betting that AI will change work the way Office did in the 1980s and 1990s ^[4]. Microsoft's pitch is that when an analyst builds that forecast next Monday, the first window she opens is Copilot ^[1]. Microsoft calls it the "OS for work" ^[1], though what it has built is narrower and easier to plan around: one app, fused from the consumer and enterprise Copilots over the past few months, with separate tabs for chat, coding and Autopilot ^[5][6]. The full power of Office is coming into Copilot for the first time ^[2]. For a rollout, that is the change with the most reach, because it puts the documents staff already open every day behind a second entry point. The audience was enterprise leaders at an invite-only event ^[1]. In March, Nadella made Jacob Andreou the single head of consumer and commercial Copilot ^[8], and Andreou made the case in consumer terms. "The bar for enterprise software is higher than it has ever been before. The expectations we have of the tools we adopt at work are set by the things we use after hours and at the weekends at home," he said ^[7]. Yet the merged app looks more like the commercial Copilot. It keeps only small nods to the consumer design Mustafa Suleyman launched two years ago ^[6][16]. Autopilot is the part an admin will be asked about first. Andreou said agents "have been almost impossible to bring into the enterprise" ^[13]. On their autonomy, he said: "It is their superpower, and it's the same thing that makes them absolutely terrifying to an IT admin" ^[14]. Nadella described the goal as control. "It can't be about leaking the core of what your firm is about," he said ^[12]. Autopilot is the cloud version of Scout, an always-on agent Microsoft announced at Build in June. According to The Verge's sources, the Scout desktop app went into maintenance mode less than six months later ^[10]. Legal concerns over Yahoo's own Scout branding were part of the reason for the rename, the same sources said ^[11]. The Verge's account does not include pricing, a release date or a description of Autopilot's admin controls. It reports that Microsoft thinks it has solved the agent problem for businesses ^[15]. It also reports that the six-month redesign caused tension between Andreou's team and Charles Lamanna's Copilot, Agents, and Platform team ^[9]. In my view, the plan to write now treats Copilot as a second entry point into Office. Pilot it where work already starts in a chat window, and hold Autopilot back until Microsoft puts in writing what it can reach. The tradeoff is timing. A pilot now trains people on an interface rebuilt within six months, and on an agent renamed within six months of its debut ^[9][10]. Waiting means writing the plan after Microsoft sets its terms, while staff keep forming habits wherever they work today. For each team, I'd use a two-by-two. One axis is whether the work starts in a document or in a conversation. The other is whether the team needs an agent that acts on its own or one that only drafts. Document-first teams that only need drafts keep Office as their main entry point, with Copilot as the add-on it has been. Teams that start in conversation and only need drafts are the pilot group for the merged app with Office inside. Where work starts in documents but the team wants autonomy, wait, because the agent has to be cleared before the interface matters.