Free · Open Source · No Signup

Learn conversational hypnotherapy. Run real sessions with an AI study partner.

A free 18-lesson course on Ericksonian and conversational hypnosis — plus an open-source AI agent that delivers any technique you've learned (induction, regression, parts work) on your own machine, with built-in safety and consent rails.

No signup · No email · No paywall · Apache 2.0 · ~$1 in API credits per session · Consent-first ethics policy

A 90-second preview

What a session actually sounds like

The agent walks every session through the same 7-phase protocol (rapport → parameters → critical-factor bypass → trancework → integration → emergence → debrief). Here's a condensed exchange from a self-work session on pre-meeting nerves.

That's PCAT in 11 turns: rapport · 4-Quadrant problem probe · intensity scale · MBL apposition ("where don't you feel it") · resource revivification · transform. Lesson 12 explains why this structure works.

Who this is for

Four kinds of people are getting real value from this

Therapists & coaches

You already work with people. Use this to add precise hypnotic language patterns to your existing toolkit — apposition, NAS, revivification — without committing to a year-long certification.

AI engineers & tinkerers

A real domain-specific agent with a structured knowledge base, safety layer, and human-in-the-loop design. Fork it, study the prompts, build your own.

People doing their own inner work

Run sessions on yourself for sleep, focus, stress, or a specific issue you keep getting stuck on. The agent is patient, never tired, and free to use as often as you want.

Students of Erickson, NLP & Clean Language

One synthesis of the lineage from Erickson through Bandler/Grinder, Grove, Gilligan, and Ledochowski — with original-voice scripts you can adapt without copyright concerns.

What's inside

Three things. All free.

Course, agent, and reference playbook — designed to work together but useful on their own.

Learn the craft properly

18 lessons walk you from your first induction to running a regression with someone in trauma — at your pace, on your phone or laptop. Plain HTML, no app or login. Each lesson is a self-contained scroll-and-go page; you can skip around and come back.

Practice with a tireless partner

The igor agent will deliver a real session on you any time — for stress, sleep, focus, or a specific issue you want to work through. Built on Claude with a 7-phase protocol and a consent-first safety layer. The agent refuses to be used on people who haven't asked for it.

Reach for the right technique fast

The included playbook is a decision tree: client says X, fire technique Y. 37 named techniques, 68 language patterns, indexed by what they do and when to use them. The reference you'll keep open in another tab during real sessions.

Getting started

Three steps to running your first session

The course and the agent are designed to be used together. You can start in either order.

Read the course

Or skim — each lesson stands alone. The material builds, but you can jump in at any part. Most people read Foundations first and then pick what's useful.

Download and let Claude install it

Grab the zip from the download page. Then open Claude, drop the zip into the chat, and ask Claude to install and run it for you. Claude knows Python, pip, virtualenvs, and how to set up your API key — it'll walk you through every step regardless of your operating system.

Run a session

When Claude finishes setup, type igor chat and run a session on yourself. Or roleplay as both operator and client. The agent walks through all seven phases and uses the techniques you've learned.

Paste this into Claude after dropping the zip in chat
I downloaded this conversational hypnotherapy agent. Please:

1. Help me unzip it and check what's inside.
2. Walk me through installing it on my [Mac / Linux / Windows]
   with a clean Python virtual environment.
3. Help me set up my Anthropic API key in the .env file.
4. Run the smoke tests (igor kb-info, igor mbl "anxious" "calm").
5. Then start my first session with igor chat.

Stop after each step so I can confirm it worked.
Try it first

One technique works without installing anything

The agent's Mind-Bending Language generator produces hypnotherapy questions for any problem/resource pair. It's pure structure — no API key, no internet, no setup. Here's what it returns for (P=anxious, R=calm):

igor mbl "anxious" "calm"
Universal MBL Formula — questions for P=anxious, R=calm

  • Where else don't you feel anxious? That's calm.
  • When else don't you feel anxious? That's calm.
  • How else are you not anxious? That's calm.
  • Who else are you with when you're not anxious? That's calm.
  • What are you doing when you're not anxious? That's calm.
  • Where beyond don't you feel anxious? That's calm.
  • When before don't you feel anxious? That's calm.
  • How totally are you not anxious right now? That's calm.

Plus 20 more variants using the 7W × 4MBL formula.
The agent picks the best 3–5 to use during a session.

This is roughly a quarter of what you'll learn in Lesson 11 (Mind-Bending Language). The generator alone is worth the download even if you never set up the API key.

The curriculum

18 lessons across 5 parts

Foundations to deep therapeutic work, ending with the agent itself. Each lesson is a standalone HTML page — no app, no login.

Why this exists

A note on what this is and isn't

I spent a long time with this material. The way Ericksonian hypnosis reframes language, emotion, and the unconscious as something you can work with rather than fight against — it changed how I communicate with other people and with myself. The problem was that getting to that point required expensive certification courses, a lot of gatekeeping, and the general sense that the good stuff was behind an NDA.

I'm a software person. The natural thing to do after understanding a system deeply is to build something that embodies it. The agent — igor — is the result of that impulse. It runs sessions through seven structured phases, uses a knowledge base of techniques drawn directly from this course, and has a safety layer that recognizes when something needs a human professional. It's open source. You can read every line, extend it, and publish your own version.

The techniques on this site are old. Milton Erickson was doing this work in the 1950s. Bandler and Grinder codified much of it in the 1970s under the name NLP. David Grove developed Clean Language in the 1990s. Stephen Gilligan, Michael Yapko, and dozens of others have refined the field since. What's newer is the synthesis — pulling these threads into a learnable sequence that a curious software person (or therapist, or coach, or just a thoughtful human) can actually absorb. This site is one synthesis among many. It isn't the canonical version and doesn't claim to be.

This is not therapy. It's training material. The agent has a safety layer that detects crisis language and escalates to professional resources rather than continuing a session. If you or someone you know needs mental health support, that's where it belongs. The agent is for self-exploration, for learning, and for practitioners who have real training to use as a tool.

If you study this, become good at it, and start working with real people — please pay it forward. Not money; just thoughtfulness. Ask for consent. Be honest about what you are and aren't. Point people toward professionals when the work goes beyond your scope. The techniques are powerful precisely because the unconscious is real and responsive. Treat that with respect.

Ethics

The four hard rules

Ethics in 30 seconds

  • Consent is non-negotiable. You don't begin until the person in front of you has clearly, freely said yes.
  • Don't hypnotize the unwilling. Curiosity and coercion are not the same thing.
  • Don't replace professional care in crises. Suicidality, psychosis, acute trauma — those belong with licensed clinicians, not with you or this agent.
  • Self-work and roleplay first; real clients only after real training. The agent is a learning tool. Graduate to working with others gradually and honestly.