// THE_PROMISEAI does the work. I teach your people to keep it: using it safely in their real daily work, sharp enough to catch its mistakes and take over when it fails, built to measure whether the time it saves builds your customers' trust.
01 · WHO_IT'S_FOR
Small teams that already feel AI arriving
Teams of 5 to 15 people: service businesses, offices, customer-service and operations teams. In a small shop the owner may hold every role at once, and the program is built for that.
Teams where people are quietly worried about what AI means for their jobs. The first session says that fear out loud instead of talking around it.
Owners who want the time AI saves to show up as better service, not as mistakes that reach a customer.
Most AI training teaches people to use the tool. Keepers teaches them to direct it and keep it: check its work, catch its mistakes, speak up, and take over when it fails.
02 · THE_ROLES
Every rank moves up one rung
Everyone gets a program name one step above their everyday role, and the frontline gets the captain's title. These are program roles, not new job titles.
RANK 1CAPTAINToday: frontline staffDirects their AI helpers, catches its mistakes, speaks up, and takes over when it fails.
RANK 2COACHToday: managersGrows Captains: coaches the checking, and approves each new use of AI.
RANK 3STEWARDToday: leaders and ownersAnswers for the whole rollout: why, how far, and at what risk.
ENVOY · from any rank Not a rank. Anyone, at any level, can be sent to carry the method to the next team, and nobody goes alone. Each new team runs its own Friction Week.
Captains direct and keep. Coaches grow Captains. Stewards answer for the whole. Envoys carry it to the next team.
03 · THE_PROGRAM
Six stages, four live sessions
One rollout for one team. It opens with the failure and ends with a measured handoff.
Leader briefingWhy now, what changes, and what waiting costs. Leaders sign a short promise to their frontline: honest misses are never punished, and anyone can pause an AI step.
Friction Week FREE SAMPLE BELOWStaff log the slow, repetitive parts of their week before any new AI. The result is the team's top three tasks for AI, each with a success measure.
Baseline checkHow many planted AI mistakes the team catches before training. Run on copies of real work, never on anything a customer sees, and reported for the team only.
Four live sessionsEach one shows a failure first, then the habit that catches it, then the law behind it. One trust habit is woven into each.
Session 1: name the fear, then Law 7 (maps are not the territory)
Session 2: Law 1 (everything costs) and Law 3 (decay)
Session 3: Law 4 (outside forces) and Law 2 (order)
Session 4: Law 5 (complexity) and Law 6 (the unknown)
Re-test and go-liveThe same check after training, so the change is measured, not assumed. Each AI task gets a go-live checklist with a named owner.
Review and handoffA 30-day and 90-day check, a review, and a handoff document so the team can run the next rollout without me.
Running alongside
QUEST TRACKPractice between sessions: spot the flaw in AI work on your own tasks. Hints, never answers.
TRUST HABITS"It works" is a claim. Sound like a person, never pretend to be one. Read, draft, approve. Whose data is it?
MAKE IT YOURSA short closing module: AI-assisted work that still sounds like the person who did it.
> Full session materials are shared with client teams only.
04 · THE_SEVEN_LAWS
The Seven Laws are the spine
Each live session teaches two of the Seven Laws of the Built World, the limits every AI build runs into. People watch a law break something in a safe setting, practice the habit that catches it, and only then learn its name.
The laws describe the machine. One rule describes the person keeping it:
The more the AI does for you, the less you practice and the less you check. So the keeping has to be practiced on purpose.
Before anyone adopts a new AI tool, the team spends one normal work week writing down what slows them down. The rule for the week: log it, don't fix it.
How it works
Listen for friction phrases in your own work and your team's.
Log each one in under a minute: the task, the phrase, how many minutes it takes, and how often it happens.
On Friday, add up the hours and pick the top three. Those become the team's first AI workflows, each with a success measure.
Phrases to listen for
YOU HEAR
IT USUALLY MEANS
"I export it and then…"
Data trapped in one tool, moved by hand
"I copy this into…"
Two systems that don't talk to each other
"I retype / re-enter…"
The same information typed twice
"Does anyone know a workaround?"
The official tool doesn't fit the real job
"This takes my whole Friday."
A big, repeating block of time
"Only one person knows how to do that."
A process living in one person's head
"I just double-check it every time."
Low trust in a step, worth a review rule
Your frontline already knows where the work breaks. Friction Week gets it on paper, so AI goes where it saves real hours.
> Never write customer names, card numbers or passwords in the log. Want a dollar figure for the hours? Try the Friction Cost Calculator.
06 · PROOF_IN_PRACTICE
Version 1, built from real work
Keepers is developed, stress-tested and now piloting with its first team. The habits it teaches are the ones I use on my own AI builds every day.
Brain Bridge: how every AI session I start is briefed from one shared, searchable memory, so decisions outlive the chat they were made in.
Case studies: three AI builds, each with the problem, how it was verified, and the result.
The pilot's numbers: hours lost to friction, team confidence with AI, and frictions turned into a first AI workflow, measured before and after. They will be published here, anonymized, including what didn't work.
Want to run Friction Week with your team, or talk about the full program?