Service
AI strategy consulting for growing companies
The hard question is not whether this technology is useful. It is which three things in your particular business are worth spending money on first, and which fashionable ideas will quietly cost you a quarter.
Why bother
Most wasted spend is a sequencing error
It rarely comes from picking the wrong technology. It comes from starting with the most visible process instead of the most tractable one, and discovering the hard part only after the budget is committed.
Volume without value
A process that looks painful but runs eleven times a month. The automation works and nobody notices.
The exception problem
A task that is genuinely standard eighty per cent of the time, where the other twenty per cent is where the cost actually lives.
Building on sand
Automating around a system already scheduled for replacement, so the work is discarded before it repays.
Method
How candidates get scored
Every process we look at is assessed on the same four axes, so the resulting order is arguable on the merits rather than on enthusiasm.
| Axis | Question it answers | Why it decides priority |
|---|---|---|
| Frequency | How often does this run, and at what volume? | Value compounds with repetition; rare tasks rarely repay the build |
| Variability | How much does each instance differ from the last? | High variability means more exception handling than automation |
| System access | Do the systems involved have usable APIs? | A closed system can turn a two-week build into a two-month one |
| Consequence of error | What happens if a step gets it wrong? | Determines how much review, logging, and approval is required |
What the engagement looks like week by week
The first week is almost entirely listening. We interview the people who run the processes under review, sit with them while they work, and collect the artefacts — spreadsheets, shared inboxes, the document everyone keeps a private copy of. What a process is supposed to do and what it actually does diverge more often than not, and the divergence is usually where the automation opportunity hides.
The second week is quantification and technical checking. We put rough volumes and cycle times against each candidate, then confirm whether the systems involved can actually be integrated: which have documented APIs, which are locked behind a vendor, which are due for replacement. A process that scores well commercially but sits on a closed system is a different project entirely, and it is better to learn that now. Where the question is narrower than the whole company — which of the processes we run today is worth automating, and in what order — a bounded workflow automation audit covers that ground on its own. Automation you already run counts as a system too: where a team has outgrown its Zap library, a migration from Zapier to n8n gets scoped as its own piece of work rather than folded into a build.
The final step is the written assessment and a working session to walk through it. You get the ranked candidates, the reasoning behind the order, effort estimates, and an explicit list of what we recommend leaving alone. If the honest answer is that one afternoon of process change would deliver more than any software, that is what the document will say.
Where governance comes in: for regulated work, or any process touching personal data, we map the assessment against the NIST AI Risk Management Framework so the roadmap arrives with its risk position already stated rather than discovered during a later security review.
Questions
What to expect from an assessment
What do we actually receive at the end?
A short written assessment naming the processes worth automating, the ones deliberately left alone, and why. Each candidate carries an estimate of effort, the systems it touches, the risks it introduces, and a rough order of value. It is meant to be usable by another firm or your own team, not only by us.
Will you recommend doing nothing?
Sometimes, and for parts of the business almost always. Plenty of processes are too low-volume, too variable, or too close to a system that is about to be replaced to be worth automating. Saying so is more useful than manufacturing a roadmap, and it is a large part of why an outside assessment is worth paying for.
Do we have to hire you to build what you recommend?
No. The assessment is deliberately written to be portable — enough detail for an in-house developer or a different vendor to act on. Many clients build the first item themselves and bring us in for the harder integration work later, which is a perfectly good outcome.
How long does an assessment take?
Usually two to three weeks for a single department, most of which is interviews and watching real work happen rather than analysis. Broader assessments take longer, but we would rather scope tightly and go deep than survey everything at a depth that cannot support a decision.
Free consultation
Tell us what is costing you the most time
Describe the process. You get a straight read on whether it is worth automating, roughly what it would take, and what we would tackle first — on a call that costs nothing.
- Under an hour, free, and no obligation
- Fixed price per stage, agreed before any work starts
- You own everything we build, documented and in your accounts
- If you do not need us, we will say so on that call
The form is the fastest way to reach us — we reply from a real person, usually within one business day.