From the first call to running on live work.
One call maps the process and establishes whether it can be automated at all. If it can, the system is then proved in four stages before it touches anything live. Every stage is gated on your approval, in writing.
Mapping the process, and whether it can be automated.
Roughly 45 minutes, and the only time commitment we ask for. It exists to answer one question honestly: is this process worth automating, and can it be done. Some calls end with the answer no.
We walk the process end to end
What starts it, what information arrives and in what form, which rules get applied and by whom, which systems it touches, and where it stalls or gets handed back. Most of the useful detail is in the exceptions, so we ask about those specifically.
We measure what it costs today
Volume, minutes per item and the loaded hourly cost of the person doing it, worked out on the call rather than estimated afterwards. That figure is agreed in writing, and everything is charged against it.
We test it against four checks
Does the work have a countable unit. Can the rules be written down. Is the data reachable by a system. Is there enough volume for 20% of the cost to be worth recovering. All four have to hold.
You get an answer either way
If it fits, a fixed price and a build plan naming what gets built and in what order. If it does not, we say so and why. The cost analysis is yours in both cases, and there is no charge for the call.
After this call nothing further is needed from your team. The build works from your existing systems and documentation.
The Proving Ladder
Once the build is done, it climbs four rungs in order. You control the gate at each one.
Tested on old work
We run it against cases you've already handled, where you already know the right answer.
It watches quietly
It works beside your team on live jobs, deciding silently. Nobody acts on it. Nothing changes for them.
You compare
A simple screen shows where it agreed with your team and where it didn't, in plain numbers.
You switch it on
It goes live in writing, on your say-so — and that's when per-item billing starts, not before.
Rungs 02 and 03 are shadow mode, and they are what keep the risk low. The system decides on every live job in parallel with your team, but nothing it decides is acted on, nothing it produces reaches a customer, and nothing is written into your systems. Your team carries on exactly as they do now and does not have to check its work. You see how often it agreed with their own decisions before you commit to anything, and per-item billing does not begin until you switch it on.
Automation projects usually fail for one of four reasons.
It was trusted too early
Switched on before anybody checked whether it matched how the business actually decides things. The ladder makes that impossible — it cannot go live until you have watched it work.
Nobody could see what it did
A system that cannot explain itself cannot be corrected. Every decision here is recorded with the reason behind it, so a disagreement becomes a fix rather than an argument.
It was never updated
Businesses change and rules drift out of date. That is why support is not optional for anything running in production.
The wrong thing was automated
The most expensive mistake is building the wrong system well. That is what the audit is for — and why some audits end with a recommendation not to build.
Common questions.
Usually two to four weeks of live running, but it is entirely your call. Some clients want a fortnight, others want a full quarter before they are comfortable. There is no pressure from me to shorten it.
That is useful information, not a problem. Every disagreement gets looked at — sometimes the rules need adjusting, and sometimes it turns out the system was right and the process was inconsistent. Both are worth knowing.
We agree it before proving starts, in writing. A common bar is that it must match your team's decision on a high proportion of cases over a set period, with no serious errors.
It adds a few weeks before you rely on it. It also removes almost all of the risk of automating a live process, which is why every build works this way.
Describe the process you want to automate.
You will receive a fixed price and a build plan.
There is no charge for the initial call and no obligation. You deal directly with the engineer who builds the system.