What an AI strategy should be
An AI strategy is a ranked list of the work AI should take on, why, in what order, and what has to be true first. A useful one names the tasks, the cost of doing them today, the data each one needs and the controls around it. Ours is written from a count of your team's work, then built.
Why most AI strategies stall
The usual AI strategy is a workshop, a list of twenty ideas and a slide ranking them by excitement. Six months later one pilot has run, nobody can say whether it worked, and the list is out of date.
Two things go wrong. The ideas come from the people in the workshop rather than from the work, so the expensive tasks on the shop floor never make the list. And the people who wrote the strategy hand it to someone else to build, who finds out what the data really looks like.
We fix both. The roadmap starts from a count of every repeating task, and the team that writes it builds the first item on it.
From audit to working system
Each stage is a fixed price. You decide at the end of each one whether to carry on.
Count the work
The Workflow Audit has every member of staff in scope document their repeating work. We price it in pounds and check the data and systems behind it.
Write the roadmap
Opportunities ranked on cost, volume, feasibility and risk. Each one says what it needs first: a data fix, a system change, a policy, or nothing.
Agree the controls
Before anything touches client data, the policy, the risk assessment and the human checkpoints are written down. See AI governance.
Build the first one
Fixed-price phases with a working demo every week, an eval harness to show it works, and logging on every automated action.
Measure and move on
We compare the task before and after, using the audit as the baseline, then pick the next item on the roadmap or stop.
What the roadmap contains
- Every repeating task found by the audit, with hours a year and cost.
- A ranked shortlist of AI opportunities, each with the reason it made the list.
- For each one: the data it depends on, the systems it touches, and what a person still checks.
- Where a plain system, a package or a process change would beat AI, said plainly.
- The governance each item needs, from a line in the AI policy to a data protection impact assessment.
- A sequence and a fixed price for the first build phase.
Picking the first project
The first AI project sets the tone for every one after it, so we choose it on boring criteria.
What makes a good first project
- It happens often, so a small saving per task adds up.
- It is well understood, so you can tell a right answer from a wrong one.
- A mistake is easy to spot and put right before it reaches a customer.
- The data it needs is already digital and in one place.
The impressive one
Customer-facing chatbots, decisions about people and anything a regulator would ask about come later, once the team trusts the logging and the review process. They are where governance has to be strongest.
Questions about AI strategy
What does an AI strategy consultant do?
An AI strategy consultant decides where AI should be used in a business, in what order, and what has to be in place first. The output is a roadmap. We also build the first items on it, which keeps the roadmap honest about what the data and systems can support.
How long does an AI strategy take?
About three weeks for the Workflow Audit, which produces the evidence, and a further week or so to write the roadmap with you. The first build phase can start straight after.
Do we have to use you to build it?
No. The audit data and the roadmap are yours and export as spreadsheets. Take them to your own team or another supplier. Most clients do ask us to build the first item, because we already know the work.
Which AI models and tools do you use?
Whichever fits the task and your data rules. That might be Microsoft Copilot you already pay for, a model from OpenAI, Anthropic or Google through your own cloud account, or a private model. We do not resell any platform.
What if the audit says AI is not worth it?
Then the report says so, and it is still yours. Sometimes the best return is a plain system or a process change, and we will tell you which.
Start with where the week goes
On a free call, tell us what your team repeats and which AI ideas are on the table. We will say whether an audit is the right first step.