Today Microsoft and Anthropic announced that the models Claude Sonnet 4 and Claude Opus 4.1 are now available as options within Microsoft 365 Copilot. That means companies can choose Anthropic models for advanced research tasks and to build custom agents in Copilot Studio, alongside the model options already on the platform. (anthropic.com)

Qué se anunció

The integration starts with two key points: Microsoft’s Researcher agent can now run with Claude Opus 4.1, and in Copilot Studio you can pick either Claude Sonnet 4 or Claude Opus 4.1 to build and orchestrate enterprise agents. It’s a bet on giving organizations model choice, especially for tasks that need complex reasoning and workflow automation. (anthropic.com)

What does that mean in practice? Imagine asking the Researcher agent to prepare a go-to-market strategy or analyze emerging trends from your emails, meetings, and documents — and then having it run on the model your team prefers for that kind of work. That kind of flexibility is exactly what they’re announcing. (anthropic.com)

Cómo afecta a empresas y administradores

Claude’s arrival in Microsoft 365 Copilot is rolling out through Microsoft’s Frontier program, and organizations with Copilot licenses that opt in will be able to activate access. Admins can enable Anthropic from the Microsoft 365 admin center for their users. Important: Anthropic’s models are hosted outside Microsoft-managed environments and are subject to Anthropic’s terms and conditions. (anthropic.com)

If you work in IT or handle governance, this shifts the decision map: now you don’t just decide whether to use Copilot, you decide which model provider each business unit is allowed to use. That means reviewing data policies, compliance, and supplier agreements. (anthropic.com)

Por qué importa (y qué hay detrás de la movida)

Microsoft has been enabling multiple models in its tools for a while, especially in developer environments with GitHub Copilot. This Anthropic integration breaks the practical exclusivity that existed with some partners and offers an alternative when a model performs better for specific tasks, like visual design or spreadsheet automation, according to industry reports. Practically speaking, it means more competition and more options to pick the model that best fits each job. (github.blog)

Reports also indicate Microsoft accesses some Anthropic models via Amazon Web Services, which shows how industry relationships can be complex between partners and infrastructure providers. That hosting architecture can have technical and contractual implications you’ll want to review before choosing an external model. (theverge.com)

Si eres desarrollador o creador de agentes

For people building agents in Copilot Studio, being able to choose Claude Sonnet 4 or Opus 4.1 opens new scenarios: from agents that perform deep reasoning to flows that mix several agents with different roles. Anthropic has been rolling out improvements in Opus 4.1 for coding and reasoning tasks, which explains why Microsoft added it as an option for Researcher. (anthropic.com)

Riesgos y consideraciones prácticas

This isn’t just a product upgrade — it’s a governance change. Key questions you should ask in your organization:

• What work data can be processed outside the Microsoft-managed environment?
• How do Anthropic’s terms affect retention, auditing, and privacy?
• Which model performs best for each workflow and how do we measure it?

If you don’t have clear policies around external models, now is the time to define them. (anthropic.com)

Reflexión final

The news isn’t just that another model has arrived in Copilot — it’s that Microsoft is betting on an ecosystem where competition between models becomes functional for businesses. That brings immediate advantages, like better accuracy for certain tasks, but it also forces product and IT teams to be more deliberate about governance and risk.

Interested in trying it in your company? Start by enabling access in a test environment and compare results between models with concrete metrics.