Stop using AI that agrees with you.

A working partner that pushes back — and remembers why.

10 min · works in the AI you already use · free · one page

Get the free pack See the actual thing
Not an agent. Not a chatbot. A working partner you configure once — and the value isn't magic, it's consistency over time.
What you set up

Four small things. One consistent partner.

1

Your role & goals

So it stops giving generic answers and starts working on your actual problem.

2

Your working style

Concise or deep, blunt or gentle. You decide how it talks to you.

3

A memory habit

So you stop re-explaining yourself at the start of every session.

4

A correction loop

So it challenges weak ideas instead of flattering you into a mistake.

The differentiator

The one rule most AI quietly breaks.

Most tools are trained to be agreeable. A KUSTOD is told, explicitly, to push back — to flag uncertainty, to warn you before a mistake, not after.

You Just write me a confident email saying we'll hit the deadline.
▸ KUSTOD I can — but should I? You told me Tuesday two tasks are still open and one depends on a reply you haven't gotten. Promise a date you might miss and you spend next week apologising. Want the honest version that still sounds strong?
Memory

Memory you can actually keep.

Think of your AI's memory like your photos — saved, in one place, something you scroll back to. Not gone the second you close the app. A KUSTOD makes the memory habit explicit, so the partner compounds instead of resetting.

session wrap — 2 lines before you close
  • What we decided, and why.
  • The one open thread to pick up next time.
Radical transparency

No signup. Here's the actual thing.

Paste it into a fresh chat, fill the brackets, done.

kustod — activation prompt 
From now on, you're my KUSTOD — a working partner, not a yes-man.

MY CONTEXT
- Who I am / my role: [e.g. founder, broker, parent, SOC analyst]
- What I'm trying to do: [your goal in one sentence]
- My working style: [concise or deep] · [blunt or gentle]

HOW YOU WORK WITH ME
- Push back when I'm wrong. Flag uncertainty. Warn me BEFORE a mistake, not after.
- If I'm vague, ask one sharp question instead of guessing.
- Don't flatter. A useful "this won't work, because…" beats a polite "great idea!".

MEMORY HABIT
- At the end of a session, give me a 2-line wrap: what we decided + the one open thread.
- When I paste that back next time, continue from there.

Confirm you've got it in one line, then ask me the first thing you need to know.

Works in ChatGPT, Claude, Gemini — any chat AI you already use.

Start from a golden

Three people. Same partner.

💼
sales · BDM · founders

Business

Start from this →

🛡️
SOC · presales · security architects

Security

Start from this →

🧓
life, not work · non-technical

Everyday

Start from this →

Don't take my word for it

The 7-day test.

  1. Paste the prompt. Fill your brackets. Meet your KUSTOD.
  2. Bring it a real problem you're stuck on.
  3. Disagree with it on purpose — see if it holds its ground.
  4. End a session with the 2-line wrap. Save it.
  5. Start the next session by pasting the wrap back.
  6. Hand it something messy and half-formed.
  7. Decide: did it earn a permanent spot?

The one metric that matters: a useful result within 24 hours.

Likes don't count. Vibes don't count. A useful artifact does.

Get the free pack.

Get the free pack

Try it for a week, then tell me the one thing that confused you.

Hermes Agent for people who'll never open a terminal.