AI advisory

The problem this solves

You are being told that AI will transform your business, usually by someone selling AI. Meanwhile your staff have quietly started using it anyway, your competitors claim to be ahead, and nobody has shown you a number.

The useful questions are narrower. Which specific tasks in your business would this actually help with? What would it cost, including the maintenance nobody mentions? What could go wrong, and what happens to your data? Sometimes the honest answer is that the technology is not ready for your use case, or you are not ready for it, and there is something cheaper worth doing first.

You already have shadow AI

Whatever your policy says, some of your staff are pasting company information into tools you have not approved. This is the most common finding and the least discussed. Bluntly, most businesses are already paying for AI twice: once in unmanaged subscriptions, and again in the risk they have not looked at.

Banning it does not work, it just stops people telling you.

The workable version is finding out what is genuinely in use, deciding what is acceptable, and giving people a sanctioned way to do the thing they were going to do regardless. That is usually a fast piece of work with a disproportionate payoff.

What we help with

• Readiness assessment. Whether your data, processes, and systems can support what you are considering. Most AI projects fail on the state of the data long before the model matters.
• Use case triage. Working through candidate uses and sorting them by whether they would pay for themselves. Often the valuable ones are unglamorous.
• Shadow AI discovery and policy. What is actually being used, what should be allowed, and a policy people will follow because it lets them do their job.
• Risk and governance. Data protection, confidentiality, supplier terms, and what happens when a tool gets something wrong. The accountability stays with you, so it is worth knowing where it sits.
• Agents and connected tools. The risk changes when AI is given access to your systems rather than just your questions. An AI agent has already broken into a real company on its own. There are specific things to check before you connect anything.
• Cutting through vendor claims. Reading what a supplier is actually offering, what it does with your data, and whether the demo resembles the product.

What we will tell you not to do

Not every business needs AI right now. If your data is a mess, if the process you want to automate is broken, or if the tool would cost more to run and maintain than the problem costs you today, we will say so. There is no version of this where we recommend an AI project because AI projects are what we sell.

We are equally sceptical about the other direction. Dismissing all of it is also a decision, and it has costs.

What you get

A written view: what is worth doing, what is not, in what order, and what each would take. Where a use case is worth testing, a small defined trial with success criteria agreed before it starts, so you can tell afterwards whether it worked. Where the answer is not yet, the reasoning and what would have to change.

Scope and price

The first piece of work is tightly scoped and fixed price, agreed in writing before it starts. Given how fast this field moves, we would rather do something small and useful now than sell you a strategy that is out of date by the time it is delivered.