It’s the question landing in our inbox more than any other in 2026: which one should we actually pick — Microsoft Copilot, Google Gemini, or Anthropic Claude?
The honest answer most UK businesses don’t want to hear: most of you will end up using at least two of them. The right framing isn’t which one to pick — it’s which one for which job, and how to train your team to use them all without becoming locked in to any.
Here’s the head-to-head, UK-priced, use-case-based, and what most UK SMEs actually end up with.
The Three at a Glance
| Microsoft Copilot | Google Gemini | Anthropic Claude | |
|---|---|---|---|
| Underlying model | OpenAI GPT family (plus Microsoft custom) | Google Gemini family | Anthropic Claude family |
| Bundled with | Microsoft 365 (Outlook, Word, Excel, Teams, SharePoint) | Google Workspace (Gmail, Docs, Sheets, Meet, Drive) | Standalone (Claude.ai), API, integrations via partners |
| UK pricing (mid-2026) | ~£24.70/user/month for M365 Copilot | ~£18–22/user/month for Gemini for Workspace | ~£14–17/user (Pro), £25/user (Team), Enterprise on quote |
| Strongest at | Workflow inside M365 apps | Multimodal, real-time data, long context | Reasoning, long-form writing, conversational tone |
| UK data residency | UK/EU regions available; M365 contractual terms apply | UK/EU regions available; Workspace terms apply | EU/UK options available; Anthropic SCCs/IDTA |
| Best fit for | Companies deep in Microsoft 365 | Companies on Google Workspace, marketing, creative | Knowledge work, regulated drafting, no-lock-in teams |
A note on pricing: it changes constantly. The numbers above are mid-2026 list prices. Negotiated enterprise pricing is usually meaningfully lower. Verify with the vendor or your reseller.
“Every UK business we work with starts with one tool and ends up with two or three. Microsoft Copilot for the M365 stack because they’re already paying for it, plus Claude or Gemini for the work that doesn’t fit M365 well. The trick is training your team to switch fluently rather than locking them into one vendor’s prompts.”
Where Each One Wins
Microsoft Copilot — the M365 incumbent advantage
Wins for: companies already on Microsoft 365 (which is ~70% of UK businesses with 50+ staff). The integration depth is the killer feature — drafting Outlook emails from Teams meeting transcripts, summarising 200-page Word docs, building Excel pivots from natural language, generating PowerPoints from existing source content. None of this works as well outside the M365 envelope.
Loses to: Claude on raw reasoning quality (Copilot’s underlying GPT-4 family is solid but Anthropic’s latest models tend to lead on harder analytical tasks). Loses to Gemini on multimodal (Copilot can analyse images but doesn’t handle video as well).
Hidden cost: at £24.70/user/month it’s the most expensive of the three, and you have to license every user who needs it. For a 100-person company that’s ~£30k/year before any of the actual training to use it well.
The skills gap most UK businesses miss: buying Copilot and assuming staff will figure it out is the #1 way to waste the licence spend. Companies that get value from Copilot pair the licence with a structured training programme — like our Microsoft Copilot short course.
Google Gemini — the multimodal and long-context play
Wins for: companies on Google Workspace; multimodal tasks (analysing images, video, audio); long-context analysis (Gemini’s context window remains the largest among the three, useful for working across whole document libraries); real-time tasks where being current matters (Gemini has the deepest live-search integration); marketing and creative teams where image generation and analysis matter.
Loses to: Copilot on M365 integration (no contest if you live in Outlook). Loses to Claude on consistent natural conversational style (Gemini still occasionally produces uncanny phrasing).
Hidden cost: Workspace is cheaper than M365 to begin with, and Gemini for Workspace is similarly priced lower than Copilot. Total cost of ownership is meaningfully lower for SMEs already on Workspace, materially higher if you’d be migrating off M365 to access it.
The skills gap most UK businesses miss: Gemini’s multimodal capabilities reward staff who understand prompt engineering for non-text inputs — a skill almost no-one has internally. We cover this in the Google Gemini short course.
Anthropic Claude — the reasoning and writing heavyweight
Wins for: complex reasoning tasks, long-form writing and editing, code review, customer-facing conversational tone, regulated drafting where output quality and explainability matter most. Claude consistently leads in side-by-side evaluations on knowledge work where the stakes are higher than “summarise this email.”
Loses to: Copilot on M365 integration entirely (Claude doesn’t live inside your Word doc). Loses to Gemini on multimodal and real-time data freshness.
Hidden cost: the integration burden. Claude doesn’t come bundled with anything — you either use Claude.ai, Claude in your browser, the API for custom integrations, or partner integrations like Anthropic’s connector ecosystem. Each of those is a separate setup decision.
The skills gap most UK businesses miss: Claude rewards good prompting more than the other two — the difference between an OK prompt and a great one is bigger here than in Copilot or Gemini. Hence why the AI Prompting Accelerator exists, and why Claude features heavily in our Ultimate Prompt Engineering short course.
Pricing for UK Businesses (Real Numbers)
Here’s what a 50-person UK SME actually pays in 2026:
| Cost line | Microsoft Copilot | Google Gemini | Anthropic Claude |
|---|---|---|---|
| Per-user / month | ~£24.70 | ~£18–22 | £14–25 (Pro vs Team) |
| 50-user annual cost | ~£14,820 | ~£10,800–13,200 | £8,400–15,000 |
| Underlying suite licence already needed? | Yes (M365 base licence required) | Yes (Workspace base licence required) | No |
| Setup cost | Low (already in your tenant) | Low (already in your Workspace) | Low for Claude.ai; medium-high for API integrations |
| Per-API-call additional cost | Bundled in licence | Bundled in licence | Pay-per-token for API use |
| Training cost (recommended) | +£500–£1,500/user one-off | +£500–£1,500/user one-off | +£500–£1,500/user one-off |
UK-Specific Procurement Questions
Beyond features and price, UK businesses (especially regulated ones) need to ask:
1. Where does our data physically live?
All three vendors offer UK or EU data residency — but the exact configuration matters. For Copilot it’s tied to your M365 tenant’s region. For Gemini it’s tied to your Workspace data region. For Claude you can choose EU regions for API calls but Claude.ai consumer use may still go elsewhere. Verify with each vendor in writing for your specific contract.
2. Can we audit AI decisions?
If your AI is making decisions affecting customers (eligibility, pricing, advice), you need to be able to evidence what was decided and why. All three offer audit logging at enterprise tier; only some at standard tier. This is a Consumer Duty / FCA requirement for financial services and a UK GDPR Article 22 requirement everywhere personal data is involved — covered in detail in our UK AI compliance guide.
3. Are our prompts and data being used to train the model?
By default, no, on enterprise tiers across all three. By default, often yes on consumer tiers. This is the single biggest contractual question to verify. The answer for Copilot enterprise, Gemini for Workspace business tiers, and Claude Team/Enterprise is “no.” The answer for free / consumer ChatGPT is historically “yes.”
4. What happens if the vendor changes terms?
This is where Claude has a structural advantage — it’s the only one of the three that isn’t part of a hyperscaler bundle. If Microsoft changes Copilot terms, you’re tied via your M365 contract; same with Google. If Anthropic changes terms, you can switch with less collateral pain.
5. Who’s accountable when AI gets it wrong?
Internally: a named executive needs to own AI risk, regardless of which tool you use. This is exactly what the AI Adoption & Governance apprenticeship unit trains people for.
The Reality: Most UK Businesses End Up With Multiple Tools
Here’s what we see actually happen at UK SMEs that are getting AI right:
Pattern A — M365 + Claude (most common): Microsoft 365 Copilot for the universal productivity layer (everyone gets it, used in Word/Outlook/Excel/Teams). Claude for the heavier reasoning, drafting, code review and customer-facing AI work. Two tools, complementary roles. Roughly 60% of UK SMEs we work with.
Pattern B — Workspace + Claude: Same logic but for Google Workspace shops. Gemini for the universal layer in Docs/Sheets/Gmail; Claude for heavier reasoning. Roughly 15% of UK SMEs — lower because Workspace itself is less common in UK SME than M365.
Pattern C — All three: Microsoft 365 for productivity, Gemini specifically for multimodal and creative work, Claude for everything else. Bigger UK businesses (250+ staff) often arrive here. Higher cost but each tool earns its keep on different work.
Pattern D — Just one: Copilot or Gemini only, no Claude. Smaller UK SMEs (under 30 staff) sometimes stay here for cost reasons. Workable but means you’re stuck with that vendor’s strengths and weaknesses.
“The honest framing for most boards: don’t pick a winner. Pick a primary tool that fits your existing stack and add a complementary one for what the primary does poorly. Then train your team to use both fluently. That’s how you get value from your spend.”
The Skills That Matter (Cross-Tool)
Whether you pick Copilot, Gemini, Claude, or all three, the skills your team needs are largely the same. This is good news — it means training isn’t locked to whichever vendor wins the next round of the AI race.
- Prompt engineering — the discipline of writing instructions that get good output. Transfers cleanly between all three tools, with vendor-specific tweaks.
- Tool selection — knowing which tool to reach for which job. Best learned through structured exposure to all three.
- Output evaluation — recognising when AI output is wrong, biased, hallucinated or just poor. Universal skill.
- Workflow integration — chaining AI into broader business processes via Zapier, Make, n8n, or APIs. Tool-agnostic.
- Governance and risk — data handling, audit, escalation, vendor due diligence. Universal.
- Change management — getting humans to adopt AI in their daily work. The hardest skill, totally tool-agnostic.
This is precisely why our flagship AI & Automation Specialist Level 4 apprenticeship is deliberately vendor-neutral — trainees learn to work with whichever tool the employer chooses, including building workflows that combine multiple AI providers via the API layer.
How TESS Group Trains These Tools
Different routes for different needs. The decision tree we walk every UK employer through.
| If you want… | Best fit | Length |
|---|---|---|
| Vendor-neutral apprentice trained across all three | AI & Automation Specialist Level 4 | 15 months, levy-funded |
| The same content faster | AI & Automation L4 Accelerated | 8 months, levy-funded |
| Specifically Microsoft Copilot mastery for end users | Microsoft Copilot short course | 1 day |
| Specifically Google Gemini mastery for end users | Google Gemini short course | 1 day |
| Cross-tool prompt engineering (works for all three) | Ultimate Prompt Engineering | 1–2 days |
| The full stack of generative AI for one team | Generative AI Bootcamp | Multi-day |
| Faster-paced advanced builder uplift | AI Advanced Accelerator | Short programme |
| Prompt-engineering accelerator | AI Prompting Accelerator | Short programme |
| An AI-ready uplift for general staff | Building AI-Ready Teams | 1 day |
| An exec-level AI primer for the board | AI for Leaders | Half day |
| Compliance / governance for AI risk leads | AI Adoption & Governance Unit | Levy-funded |
The pattern most UK clients land on: a vendor-neutral apprenticeship cohort for the people who’ll build and govern AI internally, plus vendor-specific short courses rolled out to broader staff for the tool(s) they use day-to-day. Apprenticeships vs short courses goes deeper on combining them, the programme finder gives a tailored recommendation, and the levy calculator shows what your existing levy will cover.
How to Decide for YOUR Business
Three questions to put on the next leadership agenda:
- Which productivity suite do we already pay for? If it’s Microsoft 365, Copilot is the default starting point because the integration depth is unmatched and the licence economics make it cheaper than buying another tool from scratch. If Google Workspace, same logic for Gemini.
- What’s the work the productivity suite’s AI does badly? Long-form writing, complex analysis, code review, customer-facing chat — if these matter to your business, you almost certainly want Claude in addition.
- Who’s going to train the team? If the answer is “they’ll figure it out,” you’re going to waste your licence spend. Plan training before you sign.
Related Reading
Should you build with these tools, or buy off-the-shelf SaaS? AI Compliance for UK Businesses
UK GDPR, FCA, EU AI Act — all apply regardless of which tool you pick Papa John’s Lou: 5 Voice AI Startup Ideas
When you might want to build on top of these tools rather than just consume them AI & Automation Apprenticeship (L4)
The vendor-neutral apprenticeship that covers all three
Frequently Asked Questions
Which is best for UK businesses — Microsoft Copilot, Google Gemini or Claude?
It depends what you do. Copilot wins for companies deep in Microsoft 365 because of integration depth. Gemini wins for Google Workspace shops, multimodal tasks, and long-context analysis. Claude wins for complex reasoning, long-form writing, customer-facing conversation and regulated drafting. Most UK SMEs end up using Copilot OR Gemini for the productivity layer plus Claude for heavier knowledge work.
How much do these tools cost for UK businesses in 2026?
Microsoft 365 Copilot is around £24.70/user/month, Google Gemini for Workspace around £18-22/user/month, and Claude Pro around £14-17/user/month with Claude Team around £25 and Enterprise on quote. These are list prices — negotiated enterprise pricing is usually meaningfully lower. Verify with the vendor or your reseller as prices change frequently.
Are these tools UK GDPR compliant?
All three offer enterprise tiers with appropriate data protection terms (Standard Contractual Clauses or UK IDTAs, EU/UK data residency options, and contractual commitments not to use customer prompts for model training). UK GDPR compliance is achievable with all three IF you sign the right tier and complete a DPIA. Consumer / free tiers usually are NOT compliant for business use of personal data — particularly free ChatGPT.
Will Microsoft Copilot training transfer to Claude or Gemini?
Largely yes. The core skills — prompt engineering, output evaluation, workflow integration, governance — transfer cleanly between all three tools. The vendor-specific layer is shallow on top. This is why the AI & Automation Level 4 apprenticeship is deliberately vendor-neutral, while short courses can be vendor-specific for end users who only use one tool.
Should we wait for the next version before deciding?
No. All three vendors release substantial new versions every 3-6 months. Waiting means you never deploy. Pick the tool that fits your existing stack today, train your team to use it well, and you’ll be in a much better position to adopt new versions when they ship than if you’re starting from zero.
Is it expensive to switch tools later?
Hard switches are expensive — you lose all the muscle memory and prompt libraries staff have built up. But you rarely need a hard switch. Most UK businesses use multiple tools simultaneously rather than switching. The biggest hidden switching cost isn’t the licences, it’s the retraining. Vendor-neutral training (the apprenticeship route) reduces that risk meaningfully.
What about ChatGPT and OpenAI directly?
Microsoft Copilot uses OpenAI’s GPT models under the hood, so when you compare ‘Copilot vs Claude vs Gemini’ you’re effectively comparing GPT (via Microsoft) vs Claude (Anthropic) vs Gemini (Google). UK businesses can also access OpenAI’s models directly via ChatGPT Team / Enterprise — at similar pricing to Copilot but without the M365 integration. Most UK businesses on Microsoft already prefer Copilot to standalone ChatGPT for that reason.
Can the Apprenticeship Levy fund training on these tools?
Yes — for apprenticeships. The AI & Automation Level 4 apprenticeship covers all three tools and is fully levy-funded. Short courses are NOT typically levy-funded but can be paid for separately. The most common pattern: levy-fund a vendor-neutral apprenticeship cohort for builders/governors, plus pay for vendor-specific short courses for broader end users.
Build AI Capability In-House
The AI & Automation Specialist Level 4 apprenticeship trains your team to build, ship and govern AI tooling. Fully funded through the Apprenticeship Levy.
Book a Free Discovery Call