Who I’m not the right fit for
Most consulting sites tell you who they are for. That is easy to write and tells you very little, because everybody claims the same set of ideal clients. This is the other list — the situations where I am genuinely not the right person, and what I would point you at instead.
If a consultant has never told you that you do not need them for something, you are not talking to a consultant. You are talking to somebody selling.
You have not tried the off-the-shelf tools yet
If nobody in your organisation has seriously used the current generation of AI assistants for a few weeks, hiring anyone to build you something custom is premature. You will not know what you actually want, and a lot of what you think you want is already a product you can subscribe to.
What to do instead: give it a month of genuine use across a few roles. The specific frustrations that come out of that are the best possible input to a conversation with me later, and they cost you nothing to produce.
You want a chatbot on your website
Website chat widgets are a solved, productised category. There are good ones, they are inexpensive, and they can be running this afternoon. Building one from scratch is a way to spend considerably more money for a slightly worse result.
What to do instead: buy one. Come back to me if what you actually need is for that assistant to answer from your internal documents, act inside your systems, or handle work rather than questions — that is a genuinely different problem.
You are looking for the cheapest possible build
I am one experienced engineer, and I price accordingly. If the deciding factor is day rate, there are people who will quote lower and some of them will do a decent job.
What to do instead: if budget is the binding constraint, spend it on the assessment rather than the build. Knowing which single thing is worth doing is worth more than a cheap implementation of the wrong three.
Nobody internally owns the outcome
Every AI or automation project that has gone well had someone inside the business who wanted it to work and had the authority to change how people do things. Every one that has gone badly did not.
I can build the system. I cannot make an organisation adopt something it has not decided it wants.
What to do instead: find the owner before you find the vendor. If nobody will put their name against the outcome, the project is not ready regardless of who builds it.
You want AI without changing how anything works
The value in most of this comes from a process changing — a step disappearing, a handoff going away, someone doing something different with their Tuesday. If the requirement is that AI is added and nothing else moves, the honest forecast is an expensive system that gets used for a fortnight.
What to do instead: start somewhere the process is already under review. Change is much cheaper when something was going to change anyway.
You need someone on site, full time, indefinitely
I work with a small number of clients at a time and I am not a substitute for a permanent hire. If what you need is an engineering team member for the next two years, you need to recruit one.
What to do instead: hire the role. I am occasionally useful in that situation for a defined piece of architecture or a first implementation that gives the new hire something to build on rather than a blank page.
What I am a good fit for
Businesses with real operational complexity, systems that already exist, and manual work that has grown up around the gaps between them — where someone needs to work out what is genuinely worth automating, and then actually build it.
If that sounds like you, the free review is the fastest way to find out whether there is a project here worth doing.