Claude and Microsoft Copilot: Why the Future of AI is Multi-Model

For IT directors, CIOs, AI leads, and business leaders who want to leverage AI within their organization in a way that adds value, stays secure, and fits how employees actually work. 

Many organizations recognize the same discussion. One department is experimenting with Microsoft 365 Copilot, another team prefers to work with Claude, and yet elsewhere separate AI tools are being used for analysis, text, code or customer communication. This quickly gives rise to the idea that an organization has to choose: Copilot or Claude, Microsoft or Anthropic, one platform or a collection of separate tools. 

That contradiction is becoming less and less relevant. With the arrival of Claude within the Microsoft 365 Copilot working environment, Microsoft is showing where AI is heading: towards a multi-model reality. Not one language model for all tasks, but multiple models within a familiar working environment, used based on what they are good at. 

That changes the discussion. It's no longer just about which model is "best", but about how AI becomes part of work processes, data environments and decision-making. Especially now that AI applications are increasingly touching on sensitive business information, customer data, compliance and supervision. 

At the same time, the legal context becomes more concrete. From August 2026, the EU AI Act will impose stricter obligations on high-risk applications. AI supervision is also taking further shape in the Netherlands, with sectoral supervisors and a coordinating role for, among others, the Dutch Data Protection Authority and the National Digital Infrastructure Inspectorate. 

Claude in Microsoft Copilot thus makes one development visible in particular: organizations do not have to choose between models, but they do have to make conscious choices about where, how and under what conditions AI is used. The future of AI is not a single-model model. The future is multi-model, integrated and responsibly designed. This also creates an important basis for governance: it should not depend on one specific model, but should be applicable to all language models used within the organization. 

Why the discussion is no longer about Copilot or Claude

The first phase of generative AI often revolved around individual experiments: individual licenses, personal preferences, and tools used outside of the formal IT environment. That made sense in an early adoption phase, but it is not sustainable when AI becomes a structural part of business processes. 

The next phase is all about multi-model working. Different language models are used side by side, depending on the task, the context and the desired outcome. Sometimes speed is important, sometimes depth, sometimes creativity, sometimes controllability. No model is the best in every situation. 

That's why the integration of Claude into Microsoft Copilot is more important than just an extension of functionality. It shows that large platforms are also moving towards an AI environment in which multiple models come together. For organizations already working with Claude, this doesn't mean that Copilot is irrelevant. And for organizations that build on Microsoft 365, it doesn't mean they're locked into one model. 

We see this happening more broadly, says Pascal Willemssen, License Specialist at Wortell.

"In addition to Anthropic , other models are increasingly becoming part of the broader AI suite. For example, xAI with Grok has become available within Copilot Studio and Mistral can be added within Copilot Studio, which allows organizations to realize agents and workflows based on different models. This seems to be the beginning of a development in which Copilot is increasingly becoming the interface to underlying AI technology and more and more LLMs are becoming part of that broader AI environment." 

The core will be: how do you ensure that employees can benefit from strong AI models, without data, costs, compliance and knowledge sharing being fragmented across separate tools and personal accounts? 

The real choice is not which model wins 

Claude's integration into Microsoft Copilot shows that the AI market is maturing. Organizations are less and less likely to choose one model or one vendor as an endpoint. The value lies in the smart combination of models within a work environment that is safe, manageable and recognizable for employees. 

For organizations that already work with Claude today, this is an important insight. The use of Claude does not stand in the way of a Microsoft Copilot strategy. And vice versa, an investment in Copilot does not have to mean that other models are out of the picture. 

This makes the question more practical: which AI tasks fit which model, which data can be used, how does knowledge remain available within the organization and how can AI use be prevented from becoming fragmented across separate tools, subscriptions and working methods? 

Governance is becoming increasingly important in this regard. Not as a brake on innovation, but as a condition for making multi-model working possible in a responsible way. By setting up policy, data classification, access rights, monitoring and adoption not per model but throughout the organization, you prevent each new model from leading to separate agreements, exceptions and risks. 

Wortell helps organizations to make that step: from separate AI experiments to a mature AI working environment in which Microsoft 365 Copilot, Claude and other models can come together in a safe and valuable way. Not from the conviction that one model solves everything, but from the realization that AI only creates value when technology, adoption, governance and daily work fit together well.