One client. One workflow. Eight hours a week back.
That is the whole track record, and this page is all of it — what we did, how it was measured, and the format every result after it gets reported in.
We have shipped one thing. Here it is, at the level we can stand behind.
- What repeated
- A research workflow. The same task, run the same way, every week, by someone whose hour was worth considerably more than the task was. It did not need judgment. It needed doing.
- What we built
- A system that runs the workflow end to end and hands a person the finished result to check, instead of handing them the work to do. The person stayed in the way on purpose — they approve, they do not assemble.
- What changed
- Eight hours a week went back to the business. One client, one workflow. That is the measurement and that is also the entire sample.
- What we are not claiming
- Not an average. Not a percentage. Not a projection of what the same build would return in your business. We have one of these, and dressing a sample of one up as a pattern would be the first thing we would tell you to distrust in somebody else.
The client's name and the specifics of their business are not ours to publish, so they are not on this page. When a client is willing to be named, the format below is what gets published — and it will be the format whether the result is flattering or not.
What a case study on this page will contain, agreed before we have one to write.
Fixing where the goalposts stand is only free while there is no reason to move them, so it is being done now. When there is a second engagement, this is the shape the write-up takes.
-
The before, measured before
Hours counted during the diagnostic by watching the work, not reconstructed afterwards from memory. If nobody measured the before, there is no after to report and we will not publish one.
-
The mechanism, in plain terms
What comes in, what the system decides, where the person sits in the loop, what it hands back. Not a vendor list. If the description would not survive being read aloud to the people who do the job, it is not a description.
-
The after, with a date on it
The same measurement taken again, with how long after go-live it was taken. A number from week two and a number from month six are different claims and they will be labelled as different claims.
-
The sample size, attached to the number
One client is written as one client. No median gets published until there is something to take a median of, and when there is, the n sits next to it rather than in a footnote.
-
The part that did not work
What we scoped wrong, what had to be rebuilt, what we recommended leaving alone. A case study with no failures in it is an advertisement wearing a case study’s clothes.
Nothing above is a claim about work already done. It is the standard the next write-up has to clear before it goes on this page.
Book a diagnosticNo portfolio yet. Three things you can check instead.
A skeptical owner is right to want evidence. Until there is a body of it, the honest substitutes are the ones that cost us something to offer and cost you nothing to test.
-
A diagnostic that produces something you keep
Free, 60 minutes, and you walk away with a ranked written map of where your repetitive hours go — whether or not you ever hire us. That map is checkable against your own week. It is the only demonstration we can give you that is about your business rather than ours.
Book the diagnostic -
Terms in writing before the first day
Fixed scope at a fixed price from $5,000, with running it quoted alongside it from $1,000 a month — so the whole cost of year one is on the table before you commit to any of it. AI model usage and any tooling the build needs run on accounts we hold and operate, and are billed through to you at cost — quoted before we start, never marked up. Your own systems and data stay yours.
Ask for the terms -
The uncomfortable answers, already published
What happens if you stop paying. What you are dependent on. Why the monthly figure is not optional. All of it is written down where you can read it before you talk to us, rather than discovered in month two. A firm that only publishes the good half is a firm you find out about later.
How we work
A 60-minute diagnostic, then a fixed scope. You keep the map either way.
Free, no pitch deck. We look at what repeats and tell you which of it is worth automating — including when the answer is none of it.