What should an AI partner do for your business?
A good AI partner starts with one process that costs you time every week, builds something that takes that work on, and leaves a person approving what matters.
They should not start by selling you a tool. Before you pay for a build, they should tell you in plain words what it will cost, who owns it, where your data goes and who looks after it once it is live.
What does good AI work look like in a real business?
Good AI work looks dull from the outside: a job that used to take someone’s day now takes minutes, and nobody had to change how they work. Here are three examples from our own work, with the clients’ names left out.
Investor updates that draft themselves
A property investment group wrote every investor update by hand, which took most of a day each week. Now each update is drafted from the team’s own project records, and a person edits and approves it in minutes. Nothing is sent without that approval.
Articles that publish on a schedule
A house sale company has articles researched, written and published to its website on a schedule, to topics and rules set in advance. Each one answers a question a seller would type into a search.
Missed calls that get an answer
When a call goes unanswered, the caller gets a text back straight away and a follow-up after it, so an enquiry is not lost because everyone was busy.
Each of these started with one process, not a product, and each can be explained in a sentence.
What are the eight things to judge an AI partner on?
Judge an AI partner on eight things: proof, data, ownership, process, support, price, security and plain English. Under each one is a line on how we do it at DAX Studio, as a worked example.
Proof of real results
Ask to see work that is live in a real business, with numbers that say where they came from and what period they cover. A demo proves the idea. A system someone uses every week proves the partner.
How we do it
Where a case study gives a measured result, it shows its source and its dates. See the case studies.
How they handle your data
If the system touches personal data, your partner is a processor under UK GDPR, and you need a written agreement on what they may do with it. Ask where the data is held, who can see it and whether it trains anyone’s AI models.
How we do it
Client records are stored in a database in a UK or EU data centre, backed up every day. When AI works on your data, it goes through business AI services whose terms rule out training on it, and we agree with you what the AI is never sent. Every proposal sets out where it’s hosted and which providers handle your data.
Who owns what gets built
You should own the system, the data in it and the access to it once you have paid. If leaving would mean starting again from nothing, you are renting, and you should know that before you sign.
How we do it
You own what we build once the project fee is paid. If you leave, we hand over the code, the data and the access.
Starting from the process, not the tool
A good partner asks how the work is done today before naming any technology. If the first meeting is a product demo, the answer was chosen before the question was asked.
How we do it
The first call is a free look at the process you want to fix. Then comes a Scope: what we’ll build, what it costs and how long it takes.
Support after launch
An AI system needs looking after. Your business changes, your data changes, and something that worked in March can drift by September. Ask who fixes it, and what that costs each month.
How we do it
Care is from £350 a month. We keep the system fast and improving, and report the numbers.
Clear pricing
You should know the price before the build starts, in writing, with what is and is not included. An open-ended day rate moves the risk of an overrun on to you.
How we do it
A Scope costs £750, fixed. It sets out what we’ll build, what it costs and how long it takes. The build is then a fixed price.
Security basics
The basics are not optional: encrypted connections, software kept up to date, two-factor sign-in on admin accounts, and access only for the people who need it. A partner who is vague about these will be vague when something goes wrong.
How we do it
Every build gets those basics as standard. For larger builds, the proposal sets out sign-in, roles, encryption, audit logs and backups in writing, and we will complete your security questionnaire.
Plain-English communication
If you cannot explain to a colleague what the system does after the first meeting, it has not been explained well enough. You should also know who you will be talking to once work starts.
How we do it
The people who design it build it, and you talk to them directly. You see work in progress often, not only at the end.
What should you ask an AI partner before you sign?
Ask these ten questions before you sign, and expect a straight answer to each one.
- Which process would you start with, and why that one?
- Can we see something you built that a business is using today?
- Where do the numbers in your case studies come from?
- Where will our data be held, and who can see it?
- Will you sign a data processing agreement, and is our data used to train any AI model?
- Who checks the AI’s work before it reaches a customer?
- What do we own once the project is paid for?
- What is the price, and what would make it change?
- Who looks after it after launch, and what does that cost each month?
- What happens if we want to leave?
What are the red flags when choosing an AI partner?
The clearest red flag is a partner who talks about the technology before asking how your business works. Here are seven more.
- A demo, but nothing live in a real business.
- Results with no source and no dates.
- No clear answer on where your data goes.
- A price that depends on hours, with no upper limit.
- A guaranteed result before they have seen how you work.
- No plan for who checks the AI’s work, or who fixes it when it is wrong.
- You would not own it, or leaving would mean losing everything.
None of these makes a firm dishonest. Each one means the risk sits with you.
What does an AI partner cost, and how long does it take?
The cost depends on three things: how many of your systems it has to connect to, how much of the work needs a person to approve it, and the state of your data. The timeline depends on the same three.
A small automation with clear steps costs far less than a system your whole team works in. Treat any price given before someone has looked at the process as a guess, and ask for the price and the timeline in writing before the build starts.
How we price it: every project runs in the same four stages, and the build price is fixed before it starts. AI systems start from £3,000. Software your clients or your team sign in to starts from £10,000.
| Stage | Price | What it is |
|---|---|---|
| Scan | Free | A look at the process you want to fix. |
| Scope | £750, fixed | What we’ll build, what it costs and how long it takes. |
| Build | Fixed price | You see it working early and often, and you own it. |
| Care | From £350 a month | We keep it fast and improving, and report the numbers. |
DAX Studio · daxstudio.co.uk
A one-page checklist for choosing an AI partner
Take this checklist into the meeting. A partner worth choosing lets you tick every line.
Questions businesses in Wales ask about AI partners
Each question below is answered in its first sentence.
How much does an AI partner cost in Wales?
It depends on the size of the job, so ask for a fixed price for a defined piece of work. Cost is driven by how many systems are involved, how much needs a person to approve it and the state of your data. As a guide, our own AI systems start from £3,000, after a £750 Scope that sets the price in writing.
How long does an AI project take?
It depends on what the system has to do, and nobody can say honestly before looking at the process. A date given in the first meeting is a guess. Ask for the timeline in writing before the build starts, alongside the price, and ask to see work in progress along the way, not only at the end.
Does an AI partner need to be based in Wales?
No. Most of the work can be done from anywhere, though a partner nearby is easier to sit down with while they learn how you work. What matters more is live results. If your customers use Welsh, ask to see the system working in Welsh before you commit. We are in Cardiff and work across the UK.
Is our data safe with an AI partner?
It is safe if the partner can tell you where it is held, who can see it and what the AI is allowed to do with it. Under UK GDPR you need a written agreement with anyone who processes personal data for you. Ask whether your data is used to train AI models. The answer you want is no.
Will AI replace our staff?
It should not. Good AI work takes the repetitive part of a job, like drafting, sorting and copying between systems, and leaves the judgement with your people. The strongest systems keep a person approving what goes out. If a pitch is mostly about cutting headcount, ask what happens when the system gets something wrong.
What should we automate first?
Start with one process that happens every week, follows the same steps each time and costs someone hours. Reports, routine replies and re-typing between systems are common first choices. Avoid starting where a mistake would be expensive. A small win you can measure tells you more about a partner than a big plan.
Where to start: a free systems check
A free systems check is a 15-minute call: a look at the one process you most want to fix, and how the work is done today.
If it is worth taking further, the next step is a Scope, £750 fixed: what we’ll build, what it costs and how long it takes.
