Choosing between Claude, Copilot or Gemini for UK work in 2026 is less about which model scores highest on a benchmark and more about which one already lives inside the software you use every day. All three are capable, all three cost roughly the same for an individual, and all three now sit a click away inside the documents, inboxes and code editors a British small business already pays for. Anthropic lists Claude Pro at 20 US dollars a month billed monthly, while Microsoft and Google price their assistants in pounds, and that gap in how they bill is the first clue to how differently they fit a UK workflow.
- Claude Pro is 20 US dollars per month (17 US dollars on annual billing); Max starts at 100 US dollars per month. Anthropic does not publish a GBP price.
- Microsoft 365 Copilot for business is 19.32 pounds per user per month monthly, or 13.80 pounds on annual billing until 30 June 2026, both excluding VAT.
- Google AI Pro is 18.99 pounds per month and Google AI Ultra is 79.99 pounds per month, both in pounds for UK accounts.
- Pick by where your work already lives: writing and analysis favour Claude, a Microsoft 365 shop favours Copilot, a Google Workspace shop favours Gemini.
What you will achieve by the end of this guide
The aim here is narrow and practical: leave with one paid assistant chosen, a defensible reason for the choice, and a plan for the free trial that tests it on your real work rather than a toy prompt. We are not ranking the underlying models by raw intelligence, because for the writing, summarising, drafting and spreadsheet work most UK desks actually do, the frontier difference between Claude, the GPT and Anthropic models inside Copilot, and Gemini is now small enough that integration, data handling and price decide it. If you want the model-by-model writing comparison we ran separately, our round-up of the best AI writing assistant in the UK covers the output quality angle; this guide is about which one to commit a subscription to.
Treat the decision as five checks in order: what tasks you actually do, how sensitive your data is under UK GDPR, what software stack you already run, what the real cost is in pounds, and how the trial performs. Work through them honestly and the answer usually falls out before you reach the verdict. A sole trader who writes proposals all day reaches a different answer than a 40-person firm standing on Microsoft 365, and both are right for their own reasons.
Step one: assess the tasks you actually do
Start with a week of real work, not an imagined use case. If most of your day is long-form writing, editing, research synthesis and reasoning over documents you paste in, Claude is the natural fit. Anthropic has built Claude around long, careful prose and document analysis, and its Projects feature lets you pin reference material so every chat in a workspace shares the same context. Anthropic positions Claude Pro as the everyday productivity tier with access to its strongest models, and in our checks the writing voice holds up better across a 2,000-word draft than rivals that drift into list-heavy filler.
If your work is spreadsheets, slide decks, meeting summaries and inbox triage, the question is less about the model and more about whether the assistant can reach into those files without copy and paste. That is where Copilot and Gemini pull ahead, because each is wired into a full office suite. And if your work is code, all three now have a developer story, though the centre of gravity differs: Claude leans on Claude Code, Microsoft routes through GitHub Copilot and VS Code, and Gemini sits inside Android Studio and Google Cloud. Name your top three recurring tasks before you read on, because they decide more than any feature list.
Step two: data sensitivity and what UK GDPR demands
This is the step most individuals skip and most businesses cannot afford to. The headline difference is what each provider does with your prompts. On the consumer Claude plans, Anthropic gives you the controls to opt out of having your chats used to train its models, but the business-grade promise is stronger on the Team and Enterprise tiers, where Anthropic states it does not train on your content by default. Microsoft is the most explicit for organisations: Microsoft confirms that prompts, responses and data accessed through Microsoft Graph are not used to train its foundation models, and that Microsoft 365 Copilot is a GDPR-compliant service that sits within the European Union Data Boundary.
There is a nuance UK firms in regulated sectors should not miss. Microsoft notes that the Anthropic models offered inside some Copilot experiences are out of scope for the EU Data Boundary, so if data residency is a hard requirement, you need to check which model a given Copilot feature uses. The same caution that applies to professional work applies here: if you handle client data, read our guidance for Claude use in UK solicitors’ practices and for accountants weighing Claude against their duty of confidentiality before you paste anything sensitive into any assistant. The Information Commissioner’s Office treats AI prompts containing personal data as processing like any other, so your lawful basis and your retention policy still apply.
For consumers, the practical rule is simpler but still matters. Gemini’s personalised features can draw on your wider Google activity, and the settings that govern that are not on by default in the way some users assume. We walked through exactly which toggles to check in our piece on Gemini app privacy settings for UK users, and the same discipline, knowing what is logged and for how long, should inform whichever assistant you pick.
Step three: match the assistant to your existing software stack
This is the step that quietly decides the winner for most businesses, because an assistant you have to leave your apps to use is one you will stop using by week three. If you already pay for Microsoft 365, with Word, Excel, Outlook and Teams as your daily tools, Copilot is the assistant that draws on those files directly. Microsoft describes Copilot as working across the Microsoft 365 apps and grounding its answers in your own emails, documents and chats through Microsoft Graph, which is the integration that copy-and-paste assistants cannot match. The trade-off is that the value only appears if your data already lives in that ecosystem.
If your business runs on Google Workspace, with Gmail, Docs, Sheets and Slides, the mirror-image logic applies and Gemini is the assistant that reaches into those files. Google includes Gemini in Gmail, Docs and the wider Workspace apps on its paid AI Pro tier, so the integration story is as strong as Microsoft’s for the other camp. Claude sits apart here: it is the strongest standalone assistant rather than an embedded layer, which suits people whose work is portable, prose and documents you bring to it, rather than files locked inside one suite. If you are still deciding between the underlying assistants on capability alone, our comparison of Gemini and the GPT models is a useful companion read.
There is a real-world signal worth weighing. Large UK and global firms standing on Microsoft 365 have moved Copilot out at enormous scale, and the lesson for a smaller business is the same one we drew in our look at the Accenture Copilot rollout and what UK enterprises should learn: the assistant that wins inside an organisation is usually the one that needs no change to where files already sit. That inertia is a feature, not a failing, when it keeps people actually using the tool.
Step four: the real cost in pounds, not headline numbers
Price is where UK buyers get caught, because two of the three bill in pounds and one does not. Microsoft prices 365 Copilot for business at 19.32 pounds per user per month on a monthly commitment, or 13.80 pounds per user per month on annual billing as a limited offer running to 30 June 2026, both excluding VAT. Add VAT and the monthly figure is closer to 23 pounds a head. Note that this is the business add-on that bolts onto a paid Microsoft 365 subscription, not a standalone consumer product, so the true cost is Copilot plus your existing 365 licence. We broke the increase down in our note on the Microsoft 365 Copilot UK price change taking effect on 1 July 2026.
Google is the clearest to price for an individual: Google AI Pro is 18.99 pounds per month and Google AI Ultra is 79.99 pounds per month for UK accounts, with the Pro tier carrying the Gemini-in-Workspace features most small businesses want. Claude is the odd one out, with Anthropic listing Pro at 20 US dollars a month and Max from 100 US dollars a month with no published GBP price, so UK buyers pay the dollar amount converted at the card rate plus any applicable tax. At today’s rates Claude Pro lands close to Google AI Pro in real terms, but the currency exposure means your bill can drift month to month, which matters for budgeting a fixed subscription.
| Assistant | Individual plan | UK price | MTW read |
|---|---|---|---|
| Claude Pro | Pro (Anthropic) | 20 US dollars/mo (17 on annual), no GBP price | Best standalone writing and analysis; currency drift is the catch |
| Microsoft 365 Copilot | Copilot for business add-on | 19.32 pounds/user/mo, 13.80 on annual to 30 June, ex VAT | Only worth it if you already pay for Microsoft 365 |
| Google AI Pro | AI Pro | 18.99 pounds/mo | Cleanest price and the obvious pick for a Workspace shop |
One budgeting point UK buyers under-rate: per-seat assistants like Copilot scale linearly, so a five-person team is five licences, while a single Claude Pro or Google AI Pro seat is enough for a sole trader. Cost the assistant against headcount, not against the demo.
Step five: run a free trial that actually tests your work
Every one of the three offers a way to try before you pay, and the mistake is testing them with the same clever party-trick prompt instead of your real workload. Claude has a capable free tier you can use immediately. Google AI Pro is bundled into a Google One trial. Microsoft typically routes business Copilot through a paid trial or a monthly commitment you can cancel rather than an open-ended free tier, so read the cancellation terms before you start the clock. The point of the trial is not to be impressed; it is to find where the assistant fails on the work you do.
Build a fixed test set of five tasks from your actual week: a real email you need to reply to, a document you need summarised, a spreadsheet question, a piece of writing in your own voice, and one task that touches your existing files. Run the identical five through whichever two assistants you are shortlisting. Score them on accuracy, how much editing the output needed, and whether the integration saved you the copy-and-paste step. If you want a worked example of setting up a Google assistant properly, our Gemini setup walkthrough shows the kind of methodical first run that surfaces real differences.
Keep notes as you go, because the trial is also your switching insurance. Save the prompts that worked, because they are portable if you change your mind later, and most assistants now let you export your chat history. Switching cost between these three is mostly habit, not data lock-in, so a fortnight of disciplined testing buys you a year of confidence in the choice.
Common pitfalls when choosing and switching
The most common mistake is buying the cleverest demo rather than the assistant that fits your stack, then finding the integration you needed sits behind a higher tier or a different licence. The second is paying for two assistants at once because nobody cancelled the first trial; set a calendar reminder for the day before any trial converts to paid. The third, specific to teams, is rolling out Copilot or Gemini to everyone before checking that the data the assistant can see matches each person’s existing permissions, because both surface only what a user can already open, which means messy file permissions become messy AI answers.
A fourth pitfall is treating output as finished. All three providers are explicit that responses are not guaranteed to be factual, and the duty to check sits with you, which matters most in regulated or client-facing work. If your interest is small-business use specifically, the lessons in our piece on what UK SMEs took from the Code with Claude event apply across all three tools: start with one team, one clear use case, and measure before you scale.
Where to check current prices and plans
Prices move, and two of these three changed their UK terms this year, so confirm before you commit. Check Anthropic’s own pricing page for the current Claude Pro and Max figures, the Microsoft UK store for the 365 Copilot per-seat price and whether the annual discount still runs, and the Google AI plans page for AI Pro and AI Ultra in pounds. For a free first pass with no card, Claude’s free tier and Google’s free Gemini tier both let you sample the writing quality before any trial, and Microsoft’s Copilot Chat offers a free entry point separate from the paid business add-on.
Our verdict
Our view is that the stack you already run should decide this more than the model leaderboard. If you live in Microsoft 365, buy Copilot, because the integration with Word, Excel, Outlook and Teams is worth more than any marginal model edge, and at 19.32 pounds per seat it pays back fast for people who draft and summarise all day. If you live in Google Workspace, buy Google AI Pro at 18.99 pounds, for the same reason in mirror image. If your work is portable writing, research and analysis, and you are not wedded to one suite, Claude is the assistant we would pick for output quality, accepting the currency drift that comes with a dollar-priced bill. For everyone, run the five-task trial first, check the data-handling terms against your UK GDPR obligations, and set the cancellation reminder. What would flip our call is a UK GBP price from Anthropic or a hard data-residency requirement, either of which can rule a contender in or out before features matter.
Reader discussion
Leave a comment
Comments are moderated. Keep it useful, accurate, and on topic.