Playbooks
Three playbooks. Buy the part of the work you are in.
Define is modules 01 to 03: whether the product should exist, and what the first version leaves out. Build is the system, the build order, and production readiness. Ship is evaluation, launch, and the skills, architectures, and workflow files. Each one is its own PDF.
Prices
One payment each. The file stays behind it.
Each playbook is a paid download. Payment emails a private link for that file. Buying one leaves the other two locked.
Modules 01–03
$5
Define
Should this product exist?
The one-sentence claim, the evidence that would change your mind, and the single workflow the first version has to finish.
Paid file. It arrives by email after payment.
We email a payment link. The price is $5, once.
Modules 04, 05, and 07
$10
Build
How should the system be built?
Where the model sits in the system, the order you build in, and the security, privacy, and cost checks before anyone relies on it.
Paid file. It arrives by email after payment.
We email a payment link. The price is $10, once.
Modules 06 and 08, plus the libraries
$20
Ship
Does it work well enough to launch?
The numbers that mean working, the one channel you will keep using, and the skills, architectures, and workflow files you adapt.
Paid file. It arrives by email after payment.
We email a payment link. The price is $20, once.
All three together are $35. The free scorecard and checklists live on the frameworks page.
The split
Each playbook answers one question.
Define, then Build, then Ship, if you are walking the whole method. Each one stands on its own.
01 · $5
Define. Should this product exist?
A hypothesis, a go-or-stop threshold written before you look, and an exclusions list for the first version.
02 · $10
Build. How should the system be built?
A one-page diagram with the model boundary marked, a build sequence that starts at the largest unknown, and a production checklist with a yes or a ticket on each line.
03 · $20
Ship. Does it work well enough to launch?
An evaluation set with a pass line chosen before you measure, a launch plan with one channel and two numbers, and the library files.
The rule in all three
Do not use AI where deterministic software is clearer, cheaper, faster, or safer.
Use a model where interpretation, synthesis, generation, classification, ranking, or a conversation in ordinary language is what creates the value.
Who it is for
Written for the person making the product decisions.
This is for you if you are
- Founders deciding whether a first AI product should exist
- Product leads moving a prototype toward something a customer can rely on
- Operators automating a workflow and needing to know where a person stays in it
- Technical leads choosing an architecture, and where an agent is the wrong shape
- Builders in Cursor, Claude Code, or Codex who need the product decisions around the code
Skip them if
- Anyone looking for prompts to paste without making the decisions
- Readers who want a video course and a certificate
- Teams shopping for a model comparison or a vendor shortlist
Questions