Free · Takes 3–5 days
Let us go find the money.
Book the call, and our engineers will go through your systems the way we'd go through our own. You get a written report naming every place AI can be deployed, what it saves, what it costs, and what we'd skip.
30 min
First call
3–5 days
To your report
$0
Cost to you
There is no invoice at any point in the audit, and no clause that converts into one.
The deliverable
You leave the audit with a document, not a pitch.
This is what lands in your inbox after we've been through your systems. Written for the person who has to sign off the budget.
Opportunity register
Every workflow where AI can be deployed, with the approach, the annual saving, the build cost and the payback period for each.
What we'd leave alone
The workflows where automation costs more than it returns, named explicitly, with the reasoning.
Sequenced roadmap
What to build first, second and third, ordered by payback rather than by what is technically interesting.
Integration & risk notes
Which systems we'd touch, where your data would live, what could go wrong, and how each deployment gets monitored.
| Opportunity | Saving / yr | Build cost | Payback | Verdict |
|---|---|---|---|---|
| Final QC rework loop · in-line vision defect detection | $68K | 4 mo | Build | |
| Order-status tickets · grounded support agent | $46K | 3 mo | Build | |
| Marker & cut planning · layout optimisation | $52K | 5 mo | Build | |
| Gate attendance reconciliation · face-recognition attendance | $27K | 7 mo | Build | |
| Buyer document entry · Document AI extraction | $21K | 6 mo | Build | |
| Chatbot on the public site | Not recommended: traffic too low to repay the build | |||
| Fully autonomous purchase approvals | Not recommended: risk outweighs the saved hours |
Figures shown are an illustrative sample. Your report carries numbers measured in your own systems during the audit. Yours to keep, either way.
Book the audit
Tell us where to start looking.
Four short steps. We come back within one working day with a time for the call and what we would like read-only access to.
What we actually look at
Five places the money usually hides.
We look at real workflows rather than an org chart, which is why the audit needs a few days rather than a questionnaire.
Your website & digital front door
Traffic, conversion, support load and everything a visitor has to do by hand.
The systems holding your data
ERP, CRM, WMS, accounting, and the spreadsheets doing real work.
The floor, ward or warehouse
Where physical processes happen, because that's where the biggest wins usually hide.
The queues people work through
Support tickets, claims, approvals, QC. Anything with a backlog is a candidate.
The reports nobody trusts
Every manual reconciliation is a workflow that should have been automated.
We sign your NDA, or provide ours, before we look at a single system.
The sequence
From first call to signed-off roadmap.
Four stages, and you can stop after any of them. The only thing the audit costs you is the call.
Book a call
A working call, not a sales pitch. We map how your operation actually runs, what systems hold the data, and where the cost and the frustration sit today. Outcome: shared understanding of your operation.
We explore your systems
Our engineers go into the detail: your website, your ERP, the floor, the ticket queue, the spreadsheets people actually use. We look at real workflows, not an org chart. Outcome: every automatable workflow catalogued.
You get the AI Opportunity Report
A written report: every place AI can be deployed, what each one costs to build, what it saves per year, how long it takes, and which ones we would not recommend. Yours to keep either way. Outcome: a ranked, costed deployment roadmap.
You decide what ships
If the numbers work, we forward-deploy engineers into your team and ship the highest-value item first. If they do not, you have still lost nothing. Outcome: production AI, or a clear no.
Straight answers
The questions you're actually asking.
Including the sceptical ones. Especially those. Something not covered here? Ask it on the call. There is no cost to finding out whether we're a fit.
Because it is the fastest way to prove we understand your operation better than a competitor with a deck. Most AI vendors charge for discovery and deliver a roadmap written before they saw your data. We would rather spend a week in your systems and let the analysis win the work. If the numbers do not justify a build, you have cost us a week and saved yourself a bad project.
As much or as little as you are comfortable with. A useful audit usually means read-only access to the systems that hold your operational data and screen-shared walkthroughs of the workflows with the people who run them: all of it remote, under NDA, and where personal data is involved under a data processing agreement too. A site visit is only arranged when a floor or camera deployment needs physical measurement, and that happens during the build, not the audit. Read-only is genuinely read-only.
A consultancy sells you the analysis and stops there. We give the analysis away and are only paid if you decide to build. That inverts the incentive: a consultancy is rewarded for a long discovery phase, whereas we are rewarded for finding something worth deploying quickly. It also means our engineers write the report, not a strategy team who will never have to implement it.
That is the single most common thing we hear, and it is usually half true. Part of what the audit tells you is which opportunities your data already supports, which need a few months of collection first, and which are blocked entirely. Knowing that distinction is worth having even if you build nothing this year.
Those are the sectors where we hold the deepest playbooks, alongside retail, logistics, finance and manufacturing. But the audit method is sector-agnostic. It looks at workflows, data and cost. If we do not think we can help in your sector, the first call is where we will say so rather than after you have paid for discovery.
You keep it. There is no clause that makes it ours, no expiry, and nothing stopping you from handing it to another vendor or building it in-house. We would obviously prefer to build it with you, but a report you cannot use is not a real deliverable.
For a well-scoped first deployment on data that already exists, weeks rather than quarters. Vision systems on existing camera feeds and document automation tend to move fastest. Anything requiring new sensors, new data collection or regulatory sign-off takes longer, and the report will say so plainly rather than quoting you an optimistic date.
If you read the report and do nothing, that is a normal outcome.
The report is written, costed and yours to keep, use, or hand to a different vendor entirely. No expiry, no strings, and we will not chase you about it.
Proof of concept in two weeks. Production in two months.