The real work
Getting an answer is the easy part.
Anyone can get output in seconds. The work is deciding which tasks AI belongs in, handing it the business context it does not have, treating the first result as a draft rather than a delivery, and deciding how much checking a piece of work deserves before it leaves the building.
Some of that is about catching errors. AI can deliver a wrong answer in the same polished, confident tone as a right one, and a team moving quickly will not always stop to notice.
Thanks for getting in touch. Under our 30-day return policy, you can send the item back for a full refund and we will cover return shipping.
The policy is 14 days. Nobody had told the AI, and the sentence reads perfectly.
But most of what teams are missing sits earlier than that. They point AI at the wrong tasks. They write a prompt with none of the context that would make the answer usable. They accept a first draft because editing it feels like more work than it is. Then the useful prompt disappears into one person's chat history and nobody else benefits.
I help teams get real speed from AI without handing over their judgment. That is a set of habits, and habits are trainable.
What the workshop teaches
One progression, start to finish.
The session moves in order, and each part uses what came before it.
-
A working mental model
What the tool is actually doing when it answers, in plain terms, so people can predict where it will be strong and where it will be thin. This is the part that stops the two failure modes: refusing to use it, and trusting it flat.
-
The Judgment Grid
Sort the team's real tasks by how much time AI saves and what a mistake would cost. The grid tells you where to start, where to stay out, and where a person has to remain in the loop. You are not sorting tools. You are sorting decisions.
-
The DRAFT method
A sequence the team learns by running it on their own work: give the tool the context it is missing, read what comes back as a draft, push on it, edit it, and check the parts that matter. It ends with the sign-your-name test. Would you put your name on this as it stands?
-
Verification, sized to the stakes
Checking every line is a way to give back the time you just saved. Checking nothing is how a wrong number reaches a client. The team learns to match the check to what an error would cost, and to check the seams first: the places where the tool was working from something it was never given.
-
Build-and-Break
In small groups, the team builds an AI-assisted version of one of its own recurring tasks, then attacks it until it fails. Finding the failure in the room is cheaper than finding it in front of a customer.
-
Document, then Process, then Systems
A prompt that worked once is worth writing down. A written prompt used the same way twice is a process. A process the team trusts is what a system can be built on. This is the step that keeps the value in the team instead of leaving with whoever happened to write the good prompt.
The artifact
Where the work belongs
Every task the team brings gets placed. Two questions decide it: how much time AI saves here, and what it costs when the answer is wrong.
| Down: time AI saves Across: cost of being wrong | Low cost if it is wrong | High cost if it is wrong |
|---|---|---|
| Saves real time | Use it and move on. Drafts, first passes, reformatting, thinking out loud. | Speed pays here, and a person stays in the loop. Customer replies, reports, numbers that inform a decision. This is where Build-and-Break happens. |
| Saves little time | Not worth the setup. If the prompt takes longer than the task, do the task. | Keep it human. AI adds exposure here without buying back much. |
What the team keeps
Work they made, not notes they took.
Ten things. Six of them the team builds in the room out of its own work. Four arrive from me.
Made in the room
A map of the team's own work, showing where AI saves real time and where a person stays in the loop.
A prompt for one real recurring task, with the business context built into it.
An output the team has pushed back on, edited, and checked.
The start of a shared prompt library, so a prompt that works does not stay in one person's chat history.
One process worth documenting, written down while it is still fresh.
A first step for the following week that somebody has agreed to own.
Materials and follow-up from me
Four one-page handouts, built to sit next to a keyboard rather than in a folder.
A starter pack of prompts, written the way the session teaches, so nobody is staring at a blank box on the Monday.
A recap within 48 hours, written so it can be forwarded to whoever approved the session.
A check-in at 30 days, to find out what stuck and what did not.
How the session works
A working session, not a talk.
Before the day, I ask you and a few people on the team what the work is, what they already use AI for, and which recurring tasks are worth the time. The examples get rebuilt around those answers.
There are demonstrations, and I show my own work on screen. But nobody is spending the session watching generic examples. The team's own tasks are the material, and people spend most of the time working: alone, in pairs, and as a group, finishing with Build-and-Break.
It runs in person or over video. No technical background is required and nobody installs anything. We work inside the tools your organization has already approved, and inside whatever AI policy you have, including the parts that say do not put that in there.
It is built for one team, small enough that everyone works on their own task rather than watching someone else's.
Why me
I learned this problem before AI.
In 2002, another analyst and I reviewed online orders for fraud. The company's owner built the queue we worked from. At the end of the day, the three of us went through what we had seen and worked out what the system had missed. He changed it, and the next day we tested it against another set of real transactions.
That is where the idea I teach now came from.
Sixteen years in fraud prevention and leadership followed. I made decisions with incomplete information, hired and trained analysts, built teams, developed new leaders, wrote the procedures, and introduced the technology people relied on to make sound calls consistently. Every time, the technology arrived before the rules did.
That is the room your team is standing in with AI now.
Questions
Before you send anything
Is this for beginners?
Yes, as long as people are willing to work rather than watch. No technical background is required and the method assumes none. If nobody on the team has opened an AI tool yet, say so in the form and we can talk about whether it is too early.
How long is it?
Three hours is the full session, and it is the one I would pick if the calendar allows it. There is also a 90-minute version, cut from the same session rather than written separately, which runs the same order and keeps the capstone. The one thing it does not produce is the documented process, so that goes home as homework. Choose it when time is the constraint.
Which tools do you teach?
Whichever ones you already have. The method is not tied to one assistant, and I would rather work inside the tools your organization has approved than introduce something new for a day.
Is the workshop customized?
Yes. The intake before the session is what makes the exercises yours. The structure holds. The tasks, the examples, and the grid your team fills in are specific to you.
Can it work inside our AI policy?
Yes, and it works better when there is one. Send the policy ahead and it becomes part of the session instead of something people quietly work around.
Is it available virtually?
Yes. In person or over video. The exercises work either way, with breakout rooms standing in for tables.
How many people can attend?
It is built for one team, small enough that everyone works on their own task instead of watching someone else's. Tell me who you are thinking of putting in the room and I will tell you honestly whether that works as one session.
What should the team bring?
One recurring task each, and access to whatever AI tool they normally use.
What happens after the session?
A recap within 48 hours, written so it can be forwarded to whoever approved the session. Four one-page handouts the team keeps. A check-in from me at 30 days to find out what stuck.
What does it cost?
It depends on the size of the team and whether the session is in person or remote, so I quote after we have talked. There is no package to pick from on this page.