Founder; Ex-Meta; MIT AI Safety Fellow
I spent years as a documentation engineer at Meta, serving 3,000+ developers, and the hardest part was never the writing. It was being handed a mandate with no standard behind it. Someone tells you to improve the docs before the deadline, then leaves you to define "good enough." You reach for what everyone reaches for: a longer FAQ, a legal disclaimer, a feature list. None of it tells you whether a real person could understand what your AI does and where it fails.
Every Wednesday I grade real AI companies on exactly this, in public. The EU AI Act's Article 50 transparency obligations apply from August 2, 2026. California's training data law is already in effect. Colorado's rewrite arrives January 1, 2027. The dates keep moving, and every team I audit has quietly decided that means the work can wait. The obligation underneath them has not moved, and good intentions are not something you can hand a regulator.
This course is where I hand you the instrument I built and tested. You will audit the product that worries you and grade it against a framework that holds up when an engineer pushes back. You cannot fix a gap you cannot measure. In four weeks, you will measure yours.
Learn to audit, grade, and fix any AI product's documentation, and become the person in the room who can prove what it scores.
Run the 7-Question Audit and the ADECP framework on any product and produce a letter grade that holds up when an engineer pushes back.
Leave with a repeatable method, so you can audit the next product, and the one after that, without starting from scratch.
Apply the same instrument I use to grade AI companies in public every week on the product that actually worries you.
Stop scoring on instinct. Cite the evidence for every grade, so "I think it's fine" is never your answer to leadership again
Replace vague quality debates with a number and a band your whole team can agree on and act from.
Defend any score against pushback, because it rests on what a real user sees, not on what the company says it intended.
Rewrite failing limitation statements so your warnings guide users instead of shielding lawyers, using a formula you will reuse for years.
Turn "AI may be inaccurate" into a statement that names the failure, the trigger, and what the user should do about it.
Give users warnings they can act on, which cuts the support tickets that come from people misjudging what the AI can do.
AI Doc Strategist. Engineer. Author. Community Builder. 2x Founder. Ex Meta
The Documentation Engineer or DevRel Pro: You own the AI docs and were handed no standard to hold them to.
The Founder or Product Lead: You ship AI features and suspect your documentation would not survive a regulator or a burned customer.
The Career Shifter: You are moving into tech writing, AI governance, or compliance. You want a real skill the market is about to demand
Live sessions
Learn directly from Brittney Ball in a real-time, interactive format.
Community of peers
Stay accountable and share insights with like-minded professionals.
Certificate of completion
Share your new skills with your employer or on LinkedIn.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
Live sessions
1-2 hrs / week
Projects
2 hrs / week
Async content
1 hr / week
Maven for Teams
Reimbursement
Get your company to pay
Everything L&D needs: email template, receipts, and certificate of completion.
Get reimbursedTeam discount
Learn with your teammates
Save 20%+ when 2 or more teammates enroll in the same cohort.
Save 20%+ with a teamPrivate cohort
Run a cohort for your org
A dedicated cohort with a custom schedule and curriculum, tailored to your team.
Book a private cohort$1,950
USD