AI LabHow these workshops are run
Most tool training demos features and hopes something sticks. These sessions do the opposite: they teach the people who actually have the advanced use case, whatever their title, a repeatable way to put Claude to work on real decisions, and the judgment to keep their own. Grouped by the kind of task, run as active practice, each one leaving a reusable Skill behind.
What makes them work
Teach a fluency loop, not features
People leave with a method they can run on Monday, grounded in an established AI-fluency framework, not a list of things Claude can do.
Grouped by task, not function
The same move serves many teams at once, and it builds a reusable library instead of a bespoke session per department.
Advanced use cases, not advanced users
A real use case is not the same as fluency. Plenty of capable people make a Project, then work in Chat. The mechanics live in a pre-read; the session closes the proficiency gap inside the task, not with a 101 lecture.
Active, not passive
The core of each session is diagnosing a flawed output, not watching a perfect one. Judgment is the skill being taught.
Everyone leaves with a Skill
The move, packaged and provisioned, so it's already in their workspace the next time they face that kind of work.
Right tool, right cost
Alia handles most everyday work; Claude is for the genuine gap. Sonnet before Opus. We teach when not to reach for Claude, because the usage caps are real.
The method: an AI-fluency framework, run as a loop
Every session drills four AI-fluency competencies as one loop you run on your own work, without handing over your judgment: Delegation, Description, Discernment, Diligence.
Delegation
Deciding what to hand to AI and what to keep, and which tool and model fit the task.
Description
Communicating the task clearly enough for Claude to act on it: the what, the how, and the tone.
Discernment
Judging the output with a critical eye, catching where it is wrong, thin, or bluffing.
Diligence
Using the result responsibly: verifying before you act and owning the decision.
Each D is taught with sub-skills and a daily habit, on a plain-English model of how the system works. The goal is to change how people use AI every day, not just in the room.
The three sessions
Prototyping & Creative
app-building, hackathon-style, creativeDescribe a tool, get a working, clickable prototype. No code, no engineering ticket.
Heavy Analysis
Excel models, financial modelling, analyticsA computed model and sensitivity you run by talking, including Claude working inside Excel.
Research & Large-Doc Synthesis
strategy, investments, org designHold a whole corpus in a Project, then synthesize across it, compare sources, and surface the contradictions.
The shape of every session
Three moves: a short framing, then most of the room is hands-on, and it closes with tailored Q&A.
Run it yourself
Everything needed to deliver a session: the timed run-of-show, the follow-along prompt and the weak-output to diagnose for each session, and the three fictional task briefs. Switch between the three sessions inside.