For a long time, when we talked about artificial intelligence at Microsoft, there was one name that came up again and again: OpenAI. The relationship between both companies was decisive for the takeoff of ChatGPT and also for the launch of Copilot. But the AI board is moving quickly. New models, new players and increasingly intense competition are pushing large technology companies to diversify their bets. In that context, Microsoft’s latest move is understood.
The advertisement. Microsoft has decided to integrate Anthropic technology within Copilot, the assistant that is already part of tools such as Outlook, Teams or Excel within Microsoft 365. Among the new features is Cowork, a tool based on Anthropic technology aimed at facilitating tasks within the work environment. But that’s not all: Claude’s models will also be available within the Copilot chatbot alongside the more advanced OpenAI models, thus expanding the capabilities of the assistant without depending on a single artificial intelligence provider.
From asking for something to delegating work. Microsoft explains that Cowork is designed to go a step beyond the classic model of an assistant who answers questions or writes texts. The idea is that Copilot can take care of entire tasks within Microsoft 365. When the user makes a request, the system converts it into a work plan that runs in the background. To do this, it uses data from Outlook, Teams or Excel. From there, in theory, you propose actions, ask for clarification if needed, and allow the user to review or approve each step before the changes are applied.
Some examples. Let’s imagine, for example, that we ask Copilot to review our agenda in Outlook. The system could analyze the calendar, detect conflicts between meetings and identify lower priority meetings. From there I would propose different adjustments, such as rescheduling some appointments or reserving blocks of time to focus on more important tasks. Once those suggestions are reviewed and approved, the system itself could apply the changes automatically, accepting, rejecting or rescheduling meetings and reserving blocks of time to focus on other tasks.
The strategy. As we noted above, the move also reflects how Microsoft’s AI strategy is changing. The company has maintained a very close relationship with OpenAI for years and continues to be one of its largest shareholders, with a stake close to 27% after investments of around $13 billion since 2019. However, the rise of new models and the rapid evolution of the sector are pushing large technology companies to not depend on a single technology. Incorporating Anthropic tools within Copilot points precisely in that direction: building an ecosystem capable of relying on different models depending on the task.
Platforms before models. What we are seeing with decisions like this is that the race for AI is not limited to developing increasingly advanced models. It’s also about deciding where those capabilities are going to live. In the case of Microsoft, the answer seems quite broad: the company has been integrating Copilot into more and more products and services in its ecosystem (and also external ones). For some users, this constant presence can be very useful; For others it can be somewhat invasive. But beyond these perceptions, the movement clearly shows Microsoft’s strategy.
On the whole. So this is not just about adding another technology within Copilot, but rather reinforcing the idea that Microsoft wants to turn this assistant into a meeting point for different AI capabilities within its software. Incorporating Anthropic models alongside those of OpenAI points precisely to that scenario. Rather than relying on a single technology, the company appears to be laying the groundwork for a Copilot capable of combining different solutions as the AI market continues to evolve.
Images | Microsoft
In WorldOfSoftware | The best and worst of the Internet we know has been built on anonymity. AI brings bad news
