How to Choose an AI Agency in the UK: 10 Questions to Ask
Ten questions to ask before you hire an AI agency in the UK, what a good answer sounds like, and the red flags that should end the conversation.

Choose an AI agency that starts from your business problem rather than a tool, builds on your data and accounts, tells you the running costs up front, tests with your team before launch, and stays accountable afterwards. The ten questions below separate those agencies from the ones selling a chatbot with a logo on it. Each comes with what a good answer sounds like and the red flags to listen for.
“AI agency” now covers everyone from two people reselling a chatbot platform to large consultancies building custom models. Most UK organisations need something in between: a team that can automate real work safely and make the organisation visible in AI search, without a six-figure programme.
1. What problem would you solve first?
Good answer: a specific task, with a measure. “Your team answers the same twelve membership questions by email every day. A chatbot trained on your guidance could handle most of them, and we would measure tickets per week.”
Red flag: a product pitch before they understand your operations. If they lead with the tool, the tool is what they are selling.
2. Whose data will it use, and where does it go?
Good answer: your documents, prices and policies, kept in accounts you own, sent to the model provider under business terms that exclude training on your data, with a data processing agreement and a UK GDPR record.
Red flag: vagueness about where data is processed, or “it learns from the internet”.
3. Which models do you use, and can we switch?
Good answer: named models chosen per job for accuracy, cost and data location, with the ability to swap later without a rebuild.
Red flag: locked into one proprietary platform you cannot leave.
4. What will it cost to run each month?
Good answer: a fixed build price plus a stated monthly cost for the model, hosting and any tools, with an estimate based on your volumes.
Red flag: only the build price, or a large monthly “platform fee” that hides what you are paying for.
5. What will it never do?
Good answer: a written list. It will not give legal advice, will not change a booking without confirmation, will not discuss a member’s account without verification, and will hand over to a person when unsure.
Red flag: “it can handle anything”. Systems without limits produce surprises.
6. How will you test it before customers see it?
Good answer: testing against real questions and edge cases with the staff who will live with it, plus a soft launch.
Red flag: launch first, fix later.
7. Who owns the accounts, code and prompts?
Good answer: you do, from day one, with the agency as an administrator you can remove. Every automation documented.
Red flag: everything in the agency’s accounts. If they disappear, so does your system.
8. Can you make us visible in AI search too?
AI is changing how buyers find suppliers as well as how organisations work. An agency that understands both is useful. Good answer: they can explain AEO and GEO, show you what ChatGPT and Google AI say about you today, and describe how they would measure improvement.
Red flag: “we’ll get you to number one in ChatGPT”. Nobody can promise that.
9. What have you built, and can we speak to the client?
Good answer: named case studies with measurable results and a client willing to talk. For example, our work for Ali Legal took the firm from no visits from Google AI Overviews to around 455 a month in about six months.
Red flag: only demos and logos.
10. Who will we actually deal with?
Good answer: named people who build and run the system, reachable directly.
Red flag: a sales lead now and an account manager later who has never seen the code.
A short checklist
Before you sign, make sure you have:
- One clearly defined first job, with a measure of success.
- A fixed build price and a stated monthly running cost.
- A data processing agreement and a note of where data is processed.
- A written list of what the system must never do.
- Accounts, code and prompts in your name.
- A named contact who builds and runs it.
Where we fit
We are a UK AI agency and web agency in one, working with membership bodies, professional services firms and owner-run businesses across the UK. We build AI integrations on your own data and accounts, and run AEO and GEO so the AI your buyers use can find you. Ask us any of the ten questions above. We would rather you did.
Questions about How to Choose an AI Agency in the UK: 10 Questions to Ask
What does an AI agency do?
An AI agency designs, builds and runs AI systems for other organisations, such as chatbots, voice agents and workflow automations, and increasingly helps them become visible in AI search through AEO and GEO. A good one starts from the business problem and stays responsible for the system after launch.
How much does an AI agency cost in the UK?
Prices vary widely. A single chatbot or automation is often a few days of work at a fixed price, plus a monthly running cost for the model and tools. Larger programmes usually start with a paid discovery phase. Be wary of high monthly subscriptions for tools you could run yourself.
Should we hire an AI agency or build in-house?
If AI is core to your product, build the capability in-house. If you want a handful of reliable automations and better AI search visibility, an agency is usually faster and cheaper, as long as everything runs on accounts you own so you can bring it in-house later.
Related reading
Need help with any of this?
Send the brief and we reply the same day with a price and a date.


