Scoped, dated,and delivered.
Three ways to work with us, and one rule: nothing gets built before we agree where it belongs. Everything below is scoped, priced and dated before you commit.
Three shapes.Measured first.
Diagnostic
Three to five daysKnowing what to build, before anybody builds it.
We follow one operation end to end with the people who run it. You leave with a ranked list of what to automate, what to connect, and what to leave alone.
- Days with the team doing the work
- Cost of the process as it stands
- A ranked list, with expected return
- An honest answer, including none of the above
Build
From two weeksOne workflow, ended.
The top item on that list, built and shipped to production. Fixed scope, fixed date. Fourteen days for a moderate build, a month or more for a hard one.
- Scope framed against what week one found
- Production release from day fourteen
- Handover docs, source and evaluations
- 30-day support window
Embedded
OngoingYour AI team, without the hiring.
Continuous delivery across a portfolio of workflows. We ship, measure and widen month over month, going back to the evidence each time instead of guessing.
- Rolling releases against a live list
- Shared roadmap ownership
- Observability and drift evals
- Direct line, no ticket queue
Fixed price. You see the number before work starts.
Fourteen days,accounted for.
Three days with your team, then a build scoped against what we found. Day fourteen for a moderate workflow. A month for a hard one, and you hear that on day four.
- 01Days 1 to 3
Trace
We sit with the people doing the repeating and follow the work end to end. You get where the hours go, what refuses to talk to what, and what each fix is worth.
- 02Days 3 to 4
Frame
We cut that down to the one release worth building first. Fixed scope, fixed date, written down. The rest becomes phase two, including what you should never build.
- 03Days 4 to 14
Build
Daily builds you can open, so you correct us while correcting is cheap. A moderate build ships on day fourteen. A hard one takes a month, and you hear that on day four.
- 04After the first release
Widen
The first release earns the second. We measure what it actually returned, then take the next thing on the list, which by now is the new bottleneck.
Things worthasking first.
Is fourteen days real, or is it a marketing number?
It is the floor, not the average. Fourteen days is a first production release of one scoped, moderate workflow. A genuinely hard problem is a month or more, and you hear that on day four rather than day thirteen.
We are not sure what we want built. Is that a problem?
It is the normal starting point, and the part we are actually for. Come with the problem rather than a specification. Working out where this belongs is the first three days of the engagement.
What if you decide we should not build anything?
Then that is what you hear, and it is the cheapest week you will spend this year. Sometimes the answer is a rule change, a permission fix, or deleting a report nobody reads.
We have no clean data and a lot of legacy. Are we too messy?
That is the normal case, not the exception. Most of the value comes from systems that were never designed to talk to each other. No migration required first.
Do you replace our team?
No. We remove the part of their week a machine should have been doing. The same people, doing better work, with the drudgery gone.
Who owns what you build?
You do. Source, documentation, evaluations, infrastructure. No licensing trap, no hostage architecture. You could take it in-house the day after handover.
How do you charge?
Fixed price for the first week and for each scoped release, so you know the number before work starts. Embedded runs monthly. Never hourly, because hourly rewards slowness.