Agent Playbooks
Field manuals for people who work with AI agents.
Three books. One for engineers running coding agents on production code, one for operators running AI through a real store, and one for the executives who decide what any of it is worth. All dense PDFs written for people who already know their job and want the operating manual, not the introduction.
One-time payment, not a subscription. Instant access. 60-day refund, no questions asked.
The guides
Pick the one that matches your job.
They share a method and a house style. They share almost no content, because the work is not the same work.
For software engineers · $79
The Agent Playbook for Software Engineers
Running Claude Code, Codex CLI and OpenCode on production work: context discipline, project memory files, verification the agent can run itself, tools, permissions, hooks, headless automation, and what any of it costs.
- 104 pages
- 12 chapters
- 40+ prompts
Read the sales page and the full contents
For ecommerce operators · $79
The Agent Playbook for Ecommerce Operators
Applying AI across catalog operations, listing copy, merchandising, lifecycle messaging, support triage and weekly reporting, each with the refusal rules and review gates that make it safe to run at volume.
- 101 pages
- 12 chapters
- 34 workflows
Read the sales page and the full contents
For executive leaders · $199
The Agent Playbook for Executive Leaders
Where AI actually creates value, what it costs with the date attached, who should own it, how to measure it without relying on self-reported hours, and the questions to ask your teams and your vendors.
- 12 chapters
- Sources named and dated
- Team licences
Read the sales page and the full contents
Want more than one? The two $79 playbooks bundle at $149, and all three are $299. Both bundles are on the product pages.
Read first
Read a whole chapter before you spend anything.
Each guide publishes a full chapter as a web page. There is no email gate, no popup, and nothing to dismiss. In each case it is the chapter we would point to if someone asked whether the rest is worth the money.
For engineers it is chapter nine, on verification: how to give a coding agent a check it can run, so you stop being the verification loop. For operators it is chapter three, the brand context file, which is the one document that decides whether AI copy sounds like your brand or like everyone else's.
The reasoning behind that is simple. You cannot inspect a PDF before you buy it, and asking a stranger to trust a bullet list is a poor deal. Publishing a real chapter is the only honest version of "see for yourself".
For the executive guide the free artifact is a brief rather than a chapter: three corrections to figures that are probably already in your board deck, with every source linked and dated. Also ungated.
House rules
What all three books commit to.
Specific over sweeping
Named tools, named studies, named files. Prompts are printed in full and sources are printed with their sample size and date, so you can check either yourself.
Every claim earns its place
If a paragraph does not change what you do next, it is cut. You are paying for compression, not word count.
No outcome promises
No revenue figures, no productivity multipliers, no career claims. These books teach a method; what it produces depends on your work.
Uncertainty is labelled
Where documented behaviour and observed behaviour disagree, the book says so and gives the version it was checked against.
One payment, permanently
One payment. No subscription, and free updates to the edition you bought. Buy one guide and we will offer you the other exactly once, at the bundle difference; that is the complete upsell.
Sixty days to change your mind
Email us within 60 days for a full refund. No questions, no form, no requirement to justify it.
Sixty days, no questions asked.
Email support@anodizelabs.com within 60 days of purchase and we refund you in full. Read the refund policy for the details, of which there are few.