Aren't Claude, Copilot and ChatGPT completely different products?
Less different than the marketing suggests, and here's the part that surprises most MDs. If you're using Copilot, you're already using OpenAI's technology: Copilot's default engine is the same family of OpenAI models that powers ChatGPT - OpenAI's GPT-5.6 is now the named "preferred model" in Microsoft 365 Copilot. And since June 2026, if you're using Copilot Cowork - Microsoft's agentic layer that completes multi-step work rather than just chatting - you're using Claude: at launch it runs exclusively on Anthropic's Claude models, the same models behind Anthropic's own Claude Cowork.
So the "which AI is best" debate has quietly answered itself: the platforms are converging on each other's models. Chances are, if you're using AI at all, you're already using a variety of models. Don't get lost in the sauce of picking a winner while your competitor landscape is busy adopting.
If Copilot now runs Claude, what are you actually choosing?
You're choosing the wrapper, and the wrapper decisions are real. Same Claude capability, two routes: Claude-native, with flat per-seat pricing and your own governance; or inside Microsoft 365, with your organisation's data grounding, tenant-level admin controls, a full Copilot licence as the gate and usage billed on top in credits. There's also a UK catch worth knowing: Anthropic models sit outside Microsoft's EU Data Boundary and are switched off by default for UK organisations until an admin explicitly opts in - a governance decision, not a toggle. I've written a full breakdown of what Claude inside Copilot means for your firm.
What you should not be choosing is a dependency. Which brings us to the decision that actually matters.
How do you avoid betting the business on one model?
Build everything to transfer. If you build your AI capability hard-wired to one model, you've handed your negotiating position away. If a provider doubles its prices, you have no option but to pay. If the geopolitical landscape moves and digital sovereignty becomes mandatory for your business, you're forced to rebuild everything. Build it portable - skills and workflows on open standards that both platforms now share - and you pick your infrastructure up and move it. You'll tweak areas, but you retain your buying power. That might become essential.
The analysts have spotted the same thing: Directions on Microsoft's assessment of Claude arriving in Copilot is that Microsoft hasn't eliminated lock-in - it's moved it from the model level to the platform level. And Microsoft itself is the loudest proof of the principle: after committing roughly $13 billion to OpenAI, it still hedged - investing up to $5 billion in Anthropic and building its own models in-house. If the world's biggest software company won't depend on a single model, neither should a ten-person firm in Scotland.
The practical rule: the skills are the asset, not the licence. Train your people and build your workflows on the open standard, and the capability travels with you whichever wrapper you choose.
So which should your firm pick?
Pick deliberately, then adopt properly - because the data shows the choice of tool matters far less than what you do after you choose. The UK Government's AI adoption research found 77% of AI-adopting businesses report no revenue change - they all picked something. Our own experience across programmes: the public benchmark for generative AI time savings is about 2.2 hours per week; organisations we work with see 6-10 hours per person per week. Same technology. The difference is adoption.
If your world is Microsoft 365 end to end and tenant governance is your priority, Copilot with Cowork is a credible home - govern the switch-on and watch the metered billing. If you want Claude-native, you get the same agentic capability with predictable per-seat pricing and portability by design. We specialise in the Claude capability that now sits inside both - and we support your team on whichever wrapper you choose. Decide on your stack, your data posture and your team - not on which salesperson called first - and then put your energy where the return actually lives: the training, the adoption, and the design of the workflows your team will run every day.
Frequently asked questions
Is Copilot the same as ChatGPT?
Not the same product, but the same engine by default: Microsoft 365 Copilot runs on the same OpenAI model family that powers ChatGPT, with Claude and Microsoft's own models now selectable options. Copilot adds the Microsoft 365 wrapper - your organisation's data, admin controls and licensing.
Is Claude better than Copilot?
Wrong comparison since June 2026: Copilot's own agentic product, Copilot Cowork, runs on Claude models. The real comparison is the wrapper - Microsoft 365 grounding and tenant governance versus Claude-native flexibility and flat pricing - and which your team will actually adopt.
Can we switch AI providers later without starting again?
Yes, if you build for it now. Skills and workflows built on the open standards both platforms share are structurally portable - build once, and the capability travels. Hard-wire everything to one vendor's proprietary features and switching means rebuilding.
Weighing this up for your own firm is a 30-minute conversation, not a research project - book a discovery call. Or start smaller: type out the workflow that eats your week, and you'll get it back mapped - what AI should take on, where you stay in control, free.