The Build
Thirty days. Or it is free.
One workflow, live in your own cloud and your own accounts, inside 30 days. With an eval suite, a runbook, a rollback path, alerting, and your team trained on it. Then we do the next one. This is the whole business: a thing that runs on a Tuesday when nobody is watching.
- Time
- 30 days to the first workflow in production
- Terms
- Fixed price, fixed scopeIn production in 30 days, or you do not pay
Thirty days
The clock is the offer.
The clock starts at kickoff. Miss it and you do not pay for the build.
Access
Your cloud, your repo, your keys. If we cannot get in, the clock does not start.
The unhappy path works
We build the failure handling first. The happy path is the easy half.
Your team sees it
Milestone acceptance with the people who will use it. Their feedback changes the build.
Evals and alerting live
Known-correct test set, scheduled. Alerts on failure, cost and schema drift.
End to end, both teams
Including the day the API is down. Rollback rehearsed, not assumed.
In production
Deployed, team trained, runbook handed over. Hypercare starts.
Included
Everything you keep.
- 01
The workflow, in production
In your cloud, your repo, your API accounts, your keys. Not on our infrastructure, and not on a platform you cannot leave.
- 02
An eval suite
A test set with known-correct answers that runs on a schedule, so you find out when quality drifts instead of hearing it from a customer.
- 03
Runbook and rollback
What it does, what breaks it, how to turn it off, and how to put last week's version back. Written as we build, not after.
- 04
Observability and alerting
Logs, traces, cost per run, and an alert that goes to a human when it fails or when the upstream schema changes underneath it.
- 05
Permission scoping
Every tool the agent can call, scoped to least privilege, with the destructive ones gated behind a human. Documented as a register you can audit.
- 06
Training and change management
Your team helped design it, so they already know it. We run the sessions anyway, plus one-to-one time for whoever needs it.
- 07
30 days of hypercare
After go-live we are on call for a month. Direct access, not a ticket queue.
Do this if
- You have a scoped workflow and you want it running, not studied
- You want the code and the accounts in your own name from day one
- You have been burned by a pilot that never reached production
Do not, if
- You need ten engineers. We are two people and will tell you so
- The scope is still moving. Fixed price needs a fixed edge
- You want it hosted by us and billed per seat. We do not sell that
The Build / questions
What people actually ask.
What exactly does the 30-day guarantee cover?
A scoped single workflow, running in your production environment, inside 30 days of kickoff, or you do not pay for the build. It is a shipping guarantee, not a results guarantee. Whether the output changes your numbers depends on your data and your team, and anyone who guarantees that is selling you something.
What voids it?
Two things, both stated up front: access we asked for and did not get, and scope you changed mid-flight. Neither is a gotcha. Both go in writing before we start.
Who owns the code?
You do. Code, prompts, system instructions, orchestration logic, eval sets, documentation and runbooks, in your repository, from the first commit. There is no license to revoke and nothing to hand back.
What happens when the model we built on is retired?
Model IDs get deprecated with weeks of notice, sometimes days. We pin versions, keep the eval suite so a swap is measurable rather than hopeful, and migration is included for the life of a retainer. Without a retainer it is a small fixed-price job, and we will tell you it is coming.
What does your team still do by hand?
Thirty minutes. No deck. Sometimes the answer is that you do not need us, and that is a short call.
Book thirty minutes3 engagements at a time