What an AI consultancy does
An AI consultancy tells you where AI will pay back in your business, and where it will not, then helps you get there. The useful ones start from the work rather than the technology.
Ours starts with a count. Every member of staff in scope documents the work they repeat, and we price it in pounds. That count decides the strategy: which tasks AI should take on first, which need a plain system instead, and which should be left alone. Then we build it, because a recommendation that has to survive the build is a better recommendation.
We work with UK businesses of roughly 5 to 250 staff. Most have tried ChatGPT or Copilot, have a few ideas, and want someone to tell them which ones are worth the money.
What we do
Most engagements start with the audit and move through the rest as the value becomes clear. You are never committed to the next stage before the current one has proved itself.
AI readiness audit
The Workflow Audit. Every member of staff in scope documents their own repeating work, and we price it in pounds. You get a ranked list of what AI should take on, and whether your data and systems are ready for it.
AI strategy and implementation
A roadmap built from the audit, then the build. The first use case is chosen because it repeats often, is well understood and is easy to put right if the model gets it wrong.
AI governance and policy
An AI policy your staff can follow, a register of where AI is used, and the controls UK GDPR and the FCA expect. Start with our free UK AI policy template.
AI automation
Documents read into your systems, inboxes triaged and first drafts written. Scoped from the audit, with every automated action logged and reversible.
AI agent development
Agents for multi-step work that ordinary automation cannot follow, built with evals, an audit trail and a person signing off anything that matters.
Private AI in your cloud
A private LLM deployed in your own cloud account, for firms that cannot send client data to a public AI service.
AI enablement and training
Hands-on training, a prompt library and an AI champions network, so your team uses the tools well after we leave.
AI for financial services
For FCA-regulated firms: KYC review, suitability drafting and complaint triage, with Consumer Duty and SM&CR in mind.
For FCA-regulated firms, the deeper guides are AI automation for financial services and agentic AI in financial services.
How an engagement runs
The structure is deliberately conservative. It exists to stop you spending on the wrong thing, which is where most AI budgets go.
Audit
Everyone in scope documents their own week before we recommend anything. Fixed fee, about three weeks. You get full documentation of the repeating work and a ranked list of what is worth automating. The report is yours whether or not we build.
Strategy and governance
We sequence the opportunities against value and risk, and agree the policy and controls before anything touches client data. The first use case is one that repeats often and is easy to put right if the model errs.
Build
Fixed-price phases with a working demo every week. Every system ships with an eval harness, decision logging and a rehearsed way to switch it off.
Hand over and train
Your team learns to run it, the controls are written down, and the code and cloud account are in your name. Monthly support afterwards is optional.
Why an AI consultancy that builds
Advice that has to work
The team that writes the roadmap builds the first system on it. Recommendations that cannot be built do not make it into the report.
Evidence before opinion
The strategy comes from a count of your team’s work, so the first project is chosen on cost and volume rather than on what sounded impressive in a workshop.
Independent of any vendor
We do not resell a platform. We pick the model and setup that fit the job, and we will tell you when Copilot, a package or a spreadsheet would do.
Logged and reversible
Every automated action records its input, the model version and the output, so an answer can be explained to a client or a regulator later.
No lock-in
Built on your infrastructure, with the code handed over. The test of a good engagement is that your team can run it without us.
Fixed prices
A fixed fee for the audit and a fixed price per build phase. No day rates and no open-ended retainers.
Where we work
We are based in Bristol and South Devon and work with businesses across the UK. Discovery happens on site wherever your team sits, and the build runs remotely after that.
South West
Bristol, Bath, Exeter, Plymouth and the M5 corridor. Our home patch, and where most on-site days are spent. See AI consultancy in Bristol and Devon.
London and the South East
The City, Canary Wharf, Reading, Oxford, Cambridge, Brighton and Southampton.
The North and Midlands
Manchester, Leeds, Liverpool, Sheffield, York, Newcastle, Birmingham and Nottingham.
Scotland, Wales and Northern Ireland
Edinburgh, Glasgow, Cardiff and Belfast. On-site discovery, remote build, same as everywhere else.
Questions about AI consultancy
What does an AI consultancy do?
An AI consultancy works out where AI will pay back in a business, and where it will not, then helps the business get there. That covers a readiness audit, a strategy and roadmap, governance and policy, the build itself, and training so the team can run it. Many consultancies stop at the strategy. We build what we recommend.
How are AI consultants paid?
Most AI consultants charge by the day, on a monthly retainer, or a fixed fee per project. We use fixed fees: the Workflow Audit is a fixed fee quoted on the first call, and each build phase is a fixed price quoted from the audit. We do not charge day rates, because a day rate pays for time spent rather than for the system working.
What is the difference between an AI consultancy and an AI agency?
A consultancy decides what to build and why; an agency builds what it is asked to. The risk with a consultancy is a strategy nobody implements, and with an agency a build nobody needed. We do both in one team, so the advice has to survive contact with the code.
Do we need to be ready for AI before we start?
No. Finding out whether you are ready is the first job. The audit shows which work repeats, what it costs, and whether the data it depends on is in a state AI can use. Sometimes the answer is to fix a system first, and we will say so.
How do you avoid AI pilots that never reach production?
By choosing the first use case from evidence and building it for production from the start. Most stalled pilots picked a use case because it was interesting, then had no way to measure whether it worked. Ours come from the audit, ship with an eval harness and logging, and each phase delivers something your team uses.
Should we build an in-house AI team instead?
Often the right long-term answer is a blend. We can deliver the first systems and build your team’s capability alongside, so you are not permanently outsourced. Our guide to consultancy versus an in-house team walks through the trade-offs.
Talk to an AI consultant
Tell us what your team repeats every week, or the AI idea nobody has had time to test. We will say whether it is worth doing and what the first step costs.