You can see something is wrong. You cannot get anyone to act on it.
Six weeks. You bring the system you are already building at work. You leave able to identify the production decisions most likely to block the system, and explain them in the language of the people who need to act.
What this actually teaches
The same system, walked through all five seats.
Your architecture proposal did not die because it was wrong. It died because you argued it in back-end language to a room that thinks in acceptance criteria. Six weeks, five seats, one system: yours.
QA comes third, before either build week. You define failure before you design the thing that will fail.
Six Mondays, six Wednesdays
Monday I teach. Wednesday you bring what broke.
Two working days between them, on purpose. Questions asked after two days of trying are better questions.
| Wk | Monday · 75 min · Teaching | Wednesday · 60 min · Clinic | You produce |
|---|---|---|---|
| 01 | What Product owes everyone downstreamSep 21 | Is this worth a system at all | Decision map |
| 02 | What UX has to decide before anyone buildsSep 28 | Where the human stays in control | Control map |
| 03 | What QA needs before there is anything to testOct 5 | How you will know it failed | Acceptance and failure set |
| 04 | What the back end owes the model and the front endOct 12 | Where model ends and logic begins | Architecture and contracts |
| 05 | What the front end owes the humanOct 19 | What the user sees when it breaks | Slice definition |
| 06 | Where the five collide: live teardownOct 26 | Handoff readiness | Handoff plan |
Clinic format: submit three sentences 24 hours ahead. I pick four to six for what they teach the room. Eight minutes each. No open floor, no round-robin status.
1:00 PM Central, 75 minutes. Teaching. 7.5 hours total. Recorded, so a sprint emergency does not cost you the week.
1:00 PM Central, 60 minutes. Clinic. Come when you have something stuck. Up to 6 more hours if you use all of them.
30 minutes, between November 9 and December 4. Written submission 48 hours ahead.
8 hours required. Up to 14 hours if you use every optional clinic. That is the honest number to put in front of whoever approves your time.
Who runs it
I have held three of these seats and spent decades working across all five.
I am not pretending to be deeper than your best specialist in each seat. My value is carrying the same production decision across the handoffs where those specialties collide.
Thirty-five years. Clinical trials, enterprise BI, product leadership, data warehouses, and production AI. Here is the actual record, seat by seat.
Stephen McDanielFounder, YakData
What changes for you
You stop being the person who ships things that work in the demo.
A production file for your own system
Not exercises. By Week 6 you hold a decision map, control map, acceptance set, contracts, failure matrix, and handoff plan for the thing you are already paid to build.
One sentence that names the blocker
You will be able to say which production decision is still unsupported by evidence, who owns it, and what happens if nobody makes it.
Make the case in the language of the room
Argue the same point in Product language, QA language, UX language, or architecture language, depending on who is in front of you.
Plus one private 30-minute review with me between November 9 and December 4, for the production decision that needs a private second opinion. Bring one unresolved production decision and a written summary 48 hours ahead. You leave with my written recommendation and a next action.
If you are approving the budget
Six deliverables come back to your project.
This attaches to the system your team is already responsible for. Every document your engineer produces lands on your work, not on a case study.
It costs 7.5 required hours
Six Mondays at 75 minutes. The Wednesday clinics are optional and add up to six more hours if they use them. Less than one conference, no travel, and every session is recorded so a sprint emergency does not cost the seat.
You find out early if it should stop
Week 1 asks whether the problem deserves a system at all. That is the highest-value outcome in the program.
It de-risks the pilot you cannot evaluate
If you cannot personally judge whether the AI project is production-ready, this puts someone who can in the room for six weeks, at a fraction of what that judgment costs privately.
Three seats, $7,950
Send the engineer, the PM, and the QA lead together. They come back speaking the same language about the same system, which is most of the actual problem.
What $2,950 buys
Six documents. All of them about your system.
Not access to sessions. A production file that did not exist on September 20 and does on October 28, built one week at a time on the thing you are already paid to build.
These are six documents stalled projects commonly lack. Building them exposes the decisions that are still keeping your system from production.
Founding cohort
Ten seats for the founding cohort.
Cohort 1 is capped at 10 seats so I can stay close to every project. In exchange I ask for a written assessment at the end.
Admission requirement: you must be actively building an AI or data system at work. Without one, Weeks 3 through 6 have nothing to operate on.
What this is not: not a coding course, not implementation help, no certificate. In Production Handoff, I teach and review the production decisions while you own the build. If you want YakData to define, architect, and build the first working version for your organization, that is a private Forward Deployment engagement.
Founding cohort · Team of three $7,950
- Six live teaching sessions, Mondays 1:00 PM Central
- Six live clinics, Wednesdays 1:00 PM Central
- One private 30-minute review with Stephen
- Every session recorded
- Six working documents built on your own system
- Every working template I use, yours to keep
14-day guarantee. Attend the first two sessions. If it does not fit, full refund, no argument.
Apply for Cohort 1Applications close Wednesday, September 16.
Apply
Tell me what you are building.
I read every application myself and reply within one business day. I confirm the seat before taking payment. If Cohort 1 is full, I will tell you before you pay anything.