AI workshops · by Lior Meshullam

Stop re-explaining yourself to AI every morning.

Most people use AI like a vending machine: a prompt in, an answer out, back to zero next time. These workshops fix the setup, not the prompt. You leave with AI that knows you, does your real work, and gets sharper every week. Running on your own device, not written down in a notebook.

Book a workshop See the two workshops

I don't teach this. I run on it. Around 13 years in product, several products I own and operate, and one AI setup that grows with every one of them. Everything on this page is how I actually work.

The reframe

A better prompt was never the fix. A better workspace is.

Prompting is a skill that plateaus. A workspace is a system that compounds.

Most software is at its best on day one and decays from there. This runs the other way: a little wobbly on day one, sturdier every week, because every correction becomes memory and every repeated job becomes something it does for you. The bar to solve a problem with software just hit an all-time low. A setup that compounds is how one person does ten times what they could alone. You don't buy the setup. You buy the slope.

Two workshops

Two workshops for two different audiences.

Each one stands on its own and is built for a different starting point. Pick the one that describes you.

For non-technical people

AI Basics

An onboarding to AI. From "I tried ChatGPT once" to a daily habit, for anyone who uses a phone and email every day: office and admin staff, sales, HR, teachers, small-business owners, parents. No tech background needed.

Format
A 2-hour talk, a practice pack you do at your own pace, then a 2-hour follow-up plus a 1-hour open clinic
You need
A phone and a free account. No laptop, no paid plan.
Apps
ChatGPT, Gemini or Claude. You pick one and go deep.
  • One app set up properly, with your profile and memory
  • Three real tasks from your own week, done with AI
  • One saved workflow you built yourself, on your phone
  • A habit of checking its answers, and your own line on what you never paste
Book AI Basics See the syllabus
For people who already use AI daily

The Three-Legged Stool

An onboarding to context management and skills. For people who already use AI every day and want to get a lot more out of it: a setup that stops forgetting, follows your rules and runs your recurring work. You build a Claude Code workspace with managed memory, a knowledge base and custom skills wired to your tools. No coding.

Format
One full day, or two half-days a week apart. Mostly hands-on.
You need
A laptop with Claude Code and active billing, and a GitHub account. A short checklist walks you through the setup.
Group
6-10 people, working in pairs
  • A working AI workspace set up for you, ready to use on Monday morning
  • Several custom skills you built and ran on your own recurring work
  • Real tasks done end-to-end with the full setup
  • A routine for maintaining it so it keeps improving after the workshop
Book the Stool See the syllabus

Which one is yours? If AI is something you open now and then, and your phone is where you'd try it, it's AI Basics. If AI is already part of your day and you want it to remember you, follow your rules and run your recurring work, it's the Stool. Done AI Basics and using it every day? The Stool adds a lot on top.

AI Basics · syllabus

From first chat to daily habit.

For non-technical people. No laptop, no paid plan, no jargon. Two sessions about a week apart, with practice on your own phone in between.

How it runs

2h + practice + 2h + 1h clinic
  1. Session 1 · The talk: get orientedWhat happened from 2022 to today in six beats, who makes what (and what's actually free), how it works in plain words, how to talk to it, and a gallery of real everyday uses. Live demos on tasks the room calls out. No laptops.
  2. Practice pack · at your own paceAbout an hour across the week, on your phone: your three real tasks, a photo of a bill or letter, voice, and filling in your profile. Note one thing it got wrong and one that surprised you.
  3. Session 2 · Make it yoursDebrief on what broke. Then teach it about you (profile, memory, projects, and what you never paste), save a reusable shortcut for a task you repeat, and see the assistants that can act for you and the rules for using them safely. Short "do it now on your phone" moments throughout.
  4. Open clinic · 1 hourQuestions and setup help, one by one, for anyone who wants it.
Book AI Basics Ask about AI Basics
The Three-Legged Stool · syllabus

Context management and skills, built by hand.

For people who already use AI daily and want more from it. The workspace stands on three legs. Pull one away and it tips over: you're back to a chatbot with amnesia.

BrainThe thinking. The model that reasons, writes and pushes back. A collaborator you direct, and the one leg you can swap for a better one anytime.
AgentsThe doing. A skill plus the tools to finish a real job. The work summons them; they act, then they're gone.
ContextThe knowing. Hot memory that loads before you type, and a knowledge base it reads only the chapters a task needs from.

How it runs

1 full day or 2 half-days
Session 1 · Build your stool ~4 hours
TimeSegmentWhat happens
0:00-0:30The shiftWhy most AI use plateaus, and the model that compounds instead.
0:30-1:30Leg 1 · BrainStand up your workspace and run the onboarding interview that writes its identity.Hands-on
1:30-1:40BreakSetup snags get fixed here, not mid-build.
1:40-2:55Leg 2 · AgentsBuild your first custom agent: one skill wired to one real tool, and watch it act.Hands-on
2:55-3:55Leg 3 · ContextFeed that same agent persistent knowledge, run it again, and see it get sharper.Hands-on
3:55-4:15First runPoint the whole stool at a task you brought, then sign off and watch it write today's lessons back.Hands-on

Nobody moves on until the room is green. Each leg ends with a checkpoint every participant passes before the next one starts. Between sessions, use it on real work and note what breaks. On a single full day, that gap is lunch.

Session 2 · Live on your real work ~4 hours
TimeSegmentWhat happens
0:00-0:25DebriefWhat worked, what broke. Real friction becomes fixes.Hands-on
0:25-1:05A day in the lifeWorked examples across real recurring work: inbox, meetings, reports, drafts.
1:05-2:15Build clinic IBuild an agent for a recurring task you brought.Hands-on
2:15-2:25Break
2:25-3:35Build clinic IIA multi-step job: connect a tool and chain all three legs.Hands-on
3:35-3:55Graft, don't adoptStaying ahead without chasing every release, deciding what's safe to share, and keeping it maintained.
3:55-4:10Share-backLive demos, the multiplier when colleagues each have a stool, and Q&A.
Book the Stool Ask about the Stool
The Stool in practice

What a workspace looks like on a normal Tuesday.

One task, all three legs

"Crawl the project's new funnel." It knew which project, crawled it, and recalled its last crawl, flagging what shipped, what didn't, and what's new. "Now crawl similar funnels." It knew the niche, found real competitors and built a comparison. "Recap it as a deck and send it on Slack." It already had the tools and the access. Sent in seconds.

No re-briefing. No re-explaining. It already knew.

One stool compounds. Several multiply.

I asked my setup to open a ticket for a colleague: a table of every project that needs tests run. It knew him, had access and knew the projects. 85% done in one prompt. He took that ticket into his own setup and finished the engineering himself, about 90 minutes from my ask to a real handoff.

FAQ

The questions that come up every time.

Why not just use the chat box?

The chat apps have all three legs, just stunted versions you don't control: memory you can't shape, agents locked to someone else's intent, and threads that fill up and drift. For most people that's plenty, and AI Basics is about getting the most out of it. For daily users who want control over memory and skills, a chat answers but a workspace remembers. That's what the Stool builds.

Do I need to be technical?

For AI Basics, no. It assumes a phone and email, nothing else. For the Stool, you don't need to code, but you should already use AI every day. You'll install Claude Code and type a few commands in a terminal; the checklist before the workshop walks you through it, and the setup interviews you and writes its own starting memory.

Is my data safe?

Sharing is a value-vs-risk call, and both workshops teach you to make it on purpose. In the Stool, your knowledge lives in your own private repository, not a vendor's black box. Secrets stay out of always-on memory and every tool gets the least access it needs. If your organization has a policy, it draws the line; if it doesn't, you'll leave knowing where to draw yours.

Could this stand in for me?

For repeatable execution, more every week. For judgment and taste, no. That stays your seat. The test is "would I have done it this way?" Every no is a gap you close once, and you'll notice you review less over time.

What does the AI itself cost?

AI Basics runs on free accounts; anything paid is named as paid, never assigned. The Stool needs a Claude subscription. Because it loads only the slice of context a task needs, a well-run setup uses fewer tokens than one long chat thread, so it gets cheaper as it grows, not pricier.

How much does the workshop cost?

It depends on which workshop, the format and the group size. Send a short note and I'll come back with a quote.

Format & logistics

The essentials, side by side.

TopicAI BasicsThe Three-Legged Stool
FormatTalk, self-paced practice, demo-led follow-up, open clinicHands-on build. Mostly doing, not listening.
Length2h + 2h + 1h clinic, about a week apartOne full day, or two half-days a week apart (recommended)
GroupThe talk works for a large room; the follow-up works best in a smaller group6-10 people, in pairs
WhereOn-site or on ZoomOn-site or remote
DeviceA phone and a free accountA laptop with Claude Code, a Claude subscription and a GitHub account
PrepWrite down three tasks from last week you'd rather not have doneA short checklist two days before: about 15 minutes, plus some real work to bring
LanguageHebrew or EnglishEnglish (Hebrew on request)
Who's running it
Lior Meshullam

Lior Meshullam

I'm a product leader with around 13 years across many products and business models, and the person behind Revgineer, where I build and run my own products. For the past year I've lived inside an AI setup that grew with me and now runs much of that work. Again and again it was ahead of the features the tool makers later shipped.

I don't hand you a frozen toolkit. I hand you the way of working and a seed that grows into yours.

Get in touch

Let's find the right workshop.

Want to book, or just have a question? Leave a line and I'll reply personally.

Which workshop?

I use your details only to answer this message. Prefer email? revgineer@gmail.com