The exact model we run
on every client.
From the first day of an engagement, before anything is built: the sixteen-section knowledge base, the complete intake, the prompts that draft and audit it, and the brief that turns approved knowledge into work. Plain documents and plain-text prompts that work with any model or agent.
Twelve documents. One system.
Everything from canonical identity to blockers and approvals, and the workflow that moves a project from signed to shipped: intake, build, verify, approve, brief, execute, deploy, improve.
Run it on your own agency first. Then duplicate it per client — that is exactly how Bonsai uses it.
- 00
Implementation Instructions
The order of operations, the gates, and how long each step really takes.
- 01
16-Section Knowledge Base
Identity through blockers and approvals. Editable DOCX.
- 01
AI-Ready Markdown Version
The same structure, ready to paste into any model or agent.
- 02
PDF Field Guide
How each section is used downstream: site, content, ads, chat, video.
- 03
Complete Intake
Every question the sixteen sections need, grouped for a real interview.
- 04
Knowledge Base Builder Prompt
Drafts sections 01–16 from your sources with evidence labels and a missing-items report.
- 05
QA / Verification Prompt
Seven test questions and a pass/fail/unverified table before anything ships.
- 06
Project Build Brief
One page that turns approved knowledge into an executable brief for any build.
- 07
Deployment Checklist
Facts, offers, forms, routing, tracking, mobile, approvals. Checked before release.
- 08
Universal Project Prompt
The prompt you put in front of every AI project once the knowledge exists.
- 09
Worked Example
A complete, clearly fictional business run through the whole system.
- 10
Source & Approval Register
Claim, source, evidence status, approver. The audit trail that makes the rest trustworthy.
Straight answers.
I read the white paper. What changes with the kit?
The paper is the diagnosis. The kit is the system: all sixteen sections, the complete intake, the Builder prompt that drafts from your sources with evidence labels, the QA prompt that audits the result, the Project Build Brief, the Deployment Checklist and the Universal Project Prompt. It is the difference between knowing the method and running it.
Does it work with my AI tool?
Yes. Plain documents and plain-text prompts. Claude, ChatGPT, Gemini, a custom agent, an automation platform: the knowledge base is what you paste in front of the work.
Can I use it for clients?
Yes. The licence covers your own agency and your clients' businesses; duplicate the templates per client. You may not resell or redistribute the documents themselves.
Do I need this before the assessment?
No, they are separate. But the hour with Bryan is far more useful if you have run the kit on your own agency first, because then it starts from your real facts instead of building them live.
What if it is not for me?
Reply to your receipt within 14 days and say so. You keep the white paper either way.
Want Bryan to look at your agency with you?
The assessment is an hour with Bryan, prepared in advance, that ends with a go-to-market for going agentic and your first big blocker solved.