Microsoft adds Claude Opus 5 to 365 Copilot

Microsoft has quietly raised the ceiling on what 365 Copilot can do. The addition of Anthropic's Claude Opus 5 is not a cosmetic model swap. It brings a materially different class of reasoning into apps your team already opens every morning.

On 24 July 2026, Microsoft announced that Claude Opus 5 is available inside Microsoft 365 Copilot today. Business managers and SME owners using Word, Excel, PowerPoint, and Copilot Chat now have access to one of the most capable AI models available, without paying for a separate subscription or switching tools.

What actually changed inside Copilot

Claude Opus 5 replaces a single-turn, prompt-and-respond pattern with something more capable. The model can hold a complex problem in context across many steps, work through long documents without losing track of earlier content, and produce financial models in Excel that require conditional logic rather than simple formula suggestions.

Before this update, Copilot was genuinely useful for first drafts and quick summaries. The gap appeared when a task needed multiple reasoning steps: comparing contract clauses, building a multi-variable budget, or structuring a board report from raw meeting notes. Those tasks required moving to a separate tool.

Claude Opus 5 is rolling out now to Microsoft 365 Copilot subscribers. No additional licence is required beyond an existing Copilot seat. Access extends to Copilot Chat, Word, Excel, PowerPoint, and Copilot Studio for teams building custom agents.

Why this matters for UK business teams

For business managers and SME owners, the practical shift is about where the work actually happens. Most UK teams have already invested in Microsoft 365 licences. Copilot sits on top of that. Adding Claude Opus 5 means the reasoning upgrade arrives inside the tools already open, not in a browser tab that needs logging into separately.

The time drain that matters most here is the multi-tool juggle. A manager drafting a proposal in Word, checking figures in Excel, and using a separate AI assistant to sense-check the logic is doing three context switches per task. Claude Opus 5 inside Copilot removes at least one of those switches. That adds up across a working week.

AI adoption is 80% people and culture, 20% technology. A model upgrade only delivers if the team knows how to ask for the right output. At gecco, our AI Assistants work uses the GRAFT methodology to design assistants that are built around specific roles and tasks, so the people using them get consistently useful results rather than inconsistent ones.

Three things to do before Friday

1. Open Excel and test a multi-step prompt. Ask Copilot to build a three-scenario revenue forecast using figures from a table already in the sheet. Note where the output is stronger than before.

2. Try a long-document task in Word. Paste in a lengthy supplier contract or internal report and ask Copilot to identify the five most significant commitments or risks. Claude Opus 5's extended context handling makes this noticeably more reliable than previous versions.

3. Explore Copilot Studio if your organisation has access. The addition of Claude Opus 5 as a model option in Studio means any custom agent your team has already built can be upgraded to use it. A business manager can action this in under 30 minutes with no technical support needed.

All three actions require only an existing Microsoft 365 Copilot licence. No additional spend is needed.

Honest limits worth noting

Model choice inside Copilot is still determined by Microsoft's rollout schedule. Not every tenant will see Claude Opus 5 on day one. If it is not yet visible as a model option in your Copilot settings, it is likely queued rather than unavailable.

Claude Opus 5 is a strong reasoning model, but it does not remove the need for human review on financial outputs. A well-constructed Excel model still needs a qualified person to check assumptions. The model improves speed and structure. It does not replace judgement.

There is also a prompt quality dependency. Vague instructions produce vague results regardless of which model is running underneath. Teams who invest time in learning how to brief Copilot well will see substantially better outcomes than those who do not.

Next step for UK teams weighing up this update

If your team is already inside Microsoft 365 and wondering whether this update changes what Copilot can realistically do for your specific workflows, the AI Readiness survey is a practical starting point. Completing the survey gives you access to 65+ free resources and a custom AI Readiness report built around your organisation. That report leads directly to a free 45-minute AI Readiness call to walk through the results together.

If you want support building assistants that make the most of what Claude Opus 5 now enables inside Copilot, gecco's AI Assistants work is designed to do exactly that, using the GRAFT methodology to build role-specific assistants across your Microsoft 365 environment.

Website · LinkedIn · Case Studies · Newsletter