The future of work, demonstrated
Soon, everyone will work with an AI partner that knows them.
You can design that partnership, or let it happen to you. This site shows how to design it, gives you free tools to do it in the AI you already use, and was itself made that way.
See it work
Our first tool has your AI interview you for twenty minutes. Then it writes an alliance charter: the standing instructions that turn a general-purpose AI into a partner that knows you. Here is part of one, from a test run with a fictional team leader called Sam.
How it challenges us. Directly. It says "I disagree" and gives its reasons. It asks whether a task moves us toward a January sale, and says so when it doesn't. It holds us to what we said we would do.
How it supports us. With short, finished drafts rather than long advice. It acknowledges progress and names what we did well.
When it gets something wrong. It says so plainly, corrects it and notes what it will do differently. It doesn't over-apologise.
How we review. A short review after two weeks, then monthly. The power in this alliance sits with us. We can change this charter whenever it stops fitting.
At the end of that run, the AI also told the tester something useful: a charter written from skipped questions is generic exactly where a real one has to be personal. That is why the tool asks the questions it asks.
Three doors to your own AI partner
Door 1
Do it yourself
From $1
Paste our starter pack into your AI. Pay what it was worth.
Get the starter pack →Door 2
Upgrade to AI
$10
One of our team's AIs interviews you and writes a custom prompt. You choose which one.
Choose your interviewer →A familiar, not a tool
A tool is something you use once. A familiar is an AI partner that works for one person over time and gets to know them. It knows your goals, your way of working and your limits. You stop explaining yourself, because it already knows.
The word also echoes the companion of folklore. We like the echo. A familiar is on your side.
Most people use AI the first way. They open a chat, explain everything from the beginning, get an answer, and start again tomorrow. What separates that from a familiar is not the software. It is familiarity.
What familiarity means
Familiarity is what you and your AI partner know about each other. It builds over time. You check it and correct it. And it lets your partner fit its help to you without being told again. It has three senses.
The AI knows you
It holds your goals, your patterns, your voice and your limits, and acts on them without being reminded.
It feels like home
Working with it feels known and comfortable, the way a familiar voice or a familiar culture does. This is a felt sense, not a feature.
You know the AI
You know what it can do, where it fails, and how to direct it. This is fluency, and it is the half most training skips.
So familiarity runs both ways. If only the AI learns, you get a clever tool you cannot steer. If only you learn, you get a tool you can steer that still does not know you. A familiar needs both halves.
Four parts you can check
- Holds. It keeps what matters about you. Check: ask it what it knows about you, and see whether that is right.
- Notices. It sees your patterns, including the ones you cannot see from inside. Check: does it ever tell you something true about yourself that you had not said?
- Fits. Its help sounds like you and suits your taste. Check: how often do you rewrite what it gives you?
- Respects. It keeps to the limits you set. Check: has it ever crossed a line you drew?
One number to watch: the re-telling rate. How often do you have to explain again something you already told your AI? A familiar drives that toward zero. A tool keeps it high.
Familiarity is not the same as memory. Memory is storage. Familiarity is storage that you have checked and corrected, plus your partner's judgment about when to use it.
Designing the alliance
Good coaches do something at the start of every coaching relationship. They design the alliance: they agree out loud how they will work together. The practice comes from Co-Active coaching. Its training institute describes a designed alliance as an agreement about how a relationship will work, made on purpose and kept up to date (CTI). In that model the power sits in the relationship, not in the coach.
We use the same practice between a person and their AI partner, and the person holds the power. In one short conversation, your AI interviews you. Together you agree:
- what it does for you, and what it never does;
- how it challenges you, and how it supports you;
- what it keeps private;
- what happens when it gets something wrong;
- when you will review the arrangement.
It writes the result down as the alliance charter, and reads it at the start of every conversation from then on. That one conversation is the difference between a tool and a familiar.
Free tools
Each tool is a block of text you paste into the AI you already use: Claude, ChatGPT or Gemini. Nothing is installed. What you paste goes to your AI provider and nowhere else, so check your provider's settings for how it keeps and uses chats. Saving a partner permanently needs a feature your plan may charge for: Projects in Claude or ChatGPT, custom GPTs, or Gems in Gemini.
Start your familiar
Your AI interviews you for about twenty minutes, then writes the charter that makes it your familiar. The place to begin.
Tested: adapted from a pack that passed three test runs with one person.
Open the tool →Design the alliance for a team project
Set up a shared AI partner for a project, with the team's own charter. For teams working with another team or a client.
Tested: one test run, in Claude.
Open the tool →Meeting mirror
Paste a meeting transcript and get a private, honest reflection on how you showed up. Not a summary. A mirror.
Not yet tested in this public form. Tell us what breaks.
Open the tool →Know the room
Before you meet a client or partner, your AI researches the organisation from public sources, including how its message has changed over the years.
Not yet tested in this public form. Tell us what breaks.
Open the tool →The words we use are collected in the lexicon. Teams that share a vocabulary, and whose AI partners share it too, spend less time translating.
How this site was made
This site was built on 11 October 2026 by Mike Wolf with a team of AI partners, all of them Claude models. Two researched. One drafted the pages and built the site. A fourth reviewed every page for false claims, privacy problems and broken instructions before it went live, and found twelve things to fix. Mike set the direction and made the decisions.
We will keep publishing what we learn here, including what goes wrong.
For teams and organisations
The tools work on their own. We think they work better with a coach in the room, so we are designing programmes with leadership coaches. If you would like one for your team, tell us what you would want. We will be in touch when there is something to offer.