Case study
A mentor's framework, open to any man at any hour.
David's method lived in his documents and his sessions. The assessment lets a man start on it alone, safely, and in words that sound like David.
Live
on the mentor's own domain since October 2026
5
life domains, scored from his framework
~10 min
usual time from submission to emailed report
Engagement: Custom AI assessment build
How do you turn a mentoring framework into a paid AI assessment?
David's MIRF® assessment has been live on his own domain since October 2026. A man pays, answers questions across five life domains, and gets a personal report grounded in David's framework, usually within 10 minutes. The build kept David in charge of every sensitive word, and code checks every report for safety before it is sent.
Who they were
David, a mentor in Australia who works with men aged 30 to 55 through separation, redundancy, bereavement, and role loss. His framework, MIRF®, is his own trademarked IP. Before this build, the only way into it was working with him directly.
The problem
Men in a hard season often won't book a call. David wanted them to be able to start on his framework themselves, whenever they were ready, and still get something that sounded like him instead of a generic AI.
The audience raised the stakes. A man taking this may be struggling. The report could not diagnose, could not act as therapy, and could not sell to someone in distress. Every line of safety copy needed David's sign-off.
Scoping it honestly
The build ran in phases with sign-off gates: question set and scoring first, then the report, then the pipeline. Two beta testers took the full assessment before launch. Their feedback changed the report: larger text, fewer words per page, and a check that stops any section from being cut off.
What we built
- 01
Gated checkout
A man pays through Stripe and receives a single-use link to the assessment. A consent step in David's own words comes before the first question.
- 02
Framework knowledge base
David's framework documents sit in a vector database. New articles he publishes on his site are added automatically, so the reports keep up with his thinking.
- 03
Report engine
His answers are scored across the five domains. Claude then pulls the most relevant passages from David's material and writes a personal report: where he is under strain, what still holds, and a next step for each domain. It arrives as a branded PDF by email.
- 04
Quality and safety gate
Code checks every report for tone, length, non-clinical language, and missing sections before it is sent. A failed check sends the report back to be written again. If a man's answers show real distress, the report points him to support instead of an offer.
- 05
Owned by the mentor
The assessment runs on David's own domain. Payments land in his Stripe account, and every per-use service, from the AI to the PDF engine, bills to his own accounts.
What changed
The assessment went live on David's own domain in October 2026, after he ran the full flow himself on the live site: payment, access email, assessment, and report.
Every report passes the same automatic checks before a man sees it. In production, a report that failed its first check was rewritten and passed on the second try, with nobody touching it. The framework stays David's, and the system carries it to men at any hour.
The method stays the mentor's. The system's job is to carry it, safely, to people who aren't ready to ask in person.
More case studies
- An assessment that turned years of clinical practice into a revenue stream. →
- One article in. Eleven platform-ready assets out. →
- A decade of archives, five minutes from any answer. →
- Forty platform posts a week, four platforms, effectively one person. →
- A shop page that answers in 20 seconds, even after hours. →
- From days of stalling to a publish-ready piece in a few hours. →
- A newsletter rebuilt around one promise. →
Your version of this
Every one of these started as a conversation.
Tell me what you're sitting on — expertise, archives, a content bottleneck — and I'll tell you what I'd build.
Not ready to talk yet? Start with the AI Readiness Diagnostic.