Your AI Doesn't Know You Yet.

Take a 15-minute voice or text interview about how you work under stress, what steadies you, and what makes things worse. Leave with your own human.md, a plain-text operating manual you can hand to Claude, ChatGPT, or any AI agent.

Start Free Interview→

Free 15-minute interview. $15 one-time for your full file.

Works with Claude, ChatGPT, or any AI you use

Private by defaultPortable fileWorks anywhere
Claude
ChatGPT
AI Agents
Any Tool

AI Without Context Is Just a Tool

You tell AI what you want done. But it doesn't know who you are, how you tick, or what actually matters to you.

Without human.md

"I'm feeling overwhelmed and can't focus"

Here are some tips for overwhelm:

1. Try deep breathing exercises
2. Take a short break
3. Make a to-do list
4. Practice mindfulness
5. Consider talking to someone...

Generic. Could be for anyone.

Reply using Jonny's own human.md

"I'm feeling overwhelmed and can't focus"

Notice the urge to fix this by adding more structure. When you are anxious, you tend to over-optimise, which creates more friction when life is already full with Kelly and your daughter. Check your psoas for tightness first. Instead of reorganising your day right now, head out to the ocean and get in the salt water. We can look at your tasks when you get back.

Specific. Built from Nervous System Mastery founder Jonny Miller's own file.

What if your AI knew what helps you before you had to explain it?

One file, built around how you work

You receive a single markdown file called human.md, structured to drop straight into the AIs you already use.

01

The Quick version

A 1,200-character summary at the top of your file, sized to fit straight into ChatGPT or Claude custom-instruction fields.

02

Your full manual

Detailed sections for each focus area you chose, written in your own words as clear instructions ("When they mention X, suggest Y").

03

Practices and prompts

Matched practices linked to cards on protocolcards.com, prompts to explore, and rules so your AI suggests edits as you change.

You own your human.md. Read it, edit it in any text editor, and take it to any AI.

How It Works

No new apps to learn. No platform to switch to. Just upgrade your existing setup.

01

Pick your focus areas

Select 3 to 5 topics from 23 and do a 15-minute voice or text interview with Atlas about how you operate.

02

Read your preview

Atlas compiles your answers into one human.md file. Preview the start for free, and pay $15 one-time for the full manual.

03

Plug it into your tools

Paste the Quick version into your AI's custom instructions, and upload human.md to your projects or agent workspace.

Meet Jonny and Atlas

Jonny Miller

Jonny Miller founded Nervous System Mastery, a 5-week live cohort course on working with your stress response. Over 1,500 students have taken it. He built human.md so AI agents know how to support you under stress. Jonny uses his own human.md.

Atlas

Atlas is the AI interviewer Jonny built for human.md, running on Claude. It asks one question at a time, reflects back your own words, and asks for specific examples rather than abstract descriptions. If the questions aren't working for you, you can talk about whatever is on your mind and Atlas connects it back to the areas you chose.

What the interview covers

23 focus areas across 4 categories

Choose 3 to 5 areas to explore today. The interview goes deep on your choices to get concrete examples.

Nervous System

Understand your stress patterns and build regulation skills

Stress Triggers & Warning SignsSelf-Regulation StrategiesBurnout PreventionEmotional ProcessingEnergy & BoundariesJoy Triggers & AlivenessSleep & RecoveryBody Signals & Somatic AwarenessResilience & HRV Optimization

Life Design

Align your daily life with what matters most

Core Values & DirectionDecision-Making PrinciplesCurrent Life SeasonZone of GeniusBlind Spots & Self-SabotageIdeal Weekly RhythmKey Relationships & Support

Practical

Turn insight into action

Edge ChallengesUnhealthy PatternsReview & Reflection HabitsPast Experiments & Learnings

Deeper Work

Explore what's underneath the surface

Fear InventoryInner Parts & VoicesContemplative Practice

See what a finished human.md looks like

Every human.md is written as direct instructions to your AI, using your own phrasing. Here is an excerpt from a completed file.

human.md
# human.md
> Your operating manual for AI
> Generated 10/30/2026 | human-md.com

## Quick version

Sam is a founder scaling a team from 4 to 12 people while trying for a baby with their partner — two things that simultaneously demand everything and resist being controlled. The central tension: Sam is good at effort and fixing, but both of these challenges have hard limits on that. Warning signs are jaw clenching, compulsive Slack refreshing (looking for evidence things are okay), and being short with their partner by dinner. These typically spike after 1:1s where someone is unhappy, and peak around 4pm in back-to-back meetings. What helps: running along the river, talking to their brother, movement, getting out of their head. What makes it worse: counting-based breathing, doomscrolling, being given a list of options, being told to practise self-compassion. When Sam is spiralling, suggest exactly one small thing — something doable with a camera off or while standing. Be direct. If Sam is rewriting the same Slack message for the third time, name it. Don't over-explain or offer menus of choices.

---

## Self-Regulation Strategies

Sam has strong self-knowledge here. Use it. Don't suggest things they've already ruled out.

**What works:**
- Running along the river — this is Sam's primary reset. Protect it. If Sam hasn't mentioned a run in a few days, it's worth asking.
- Talking to their brother — co-regulation through a trusted, specific person. Not a generic "talk to someone."
- Movement and getting out of the head — the body leads recovery for Sam, not the mind.

**What doesn't work (do not suggest these):**
- Box breathing or any counting-based practice — makes Sam feel controlled and more wound up
- 10-minute structured practices — won't happen in context
- App-based tools
- Cold showers at work
- Lists of options — Sam wants one thing, not a menu
- Self-compassion language — Sam named this explicitly as unhelpful

**In-the-moment reality (4pm, camera-optional, 60 seconds or less):**
- [Physiological Sigh](https://protocolcards.com/#physiological-sigh) — two quick inhales through the nose, long exhale. No counting. Body-led.
- [Grounding Belly Sigh](https://protocolcards.com/#grounding-belly-sigh) — standing, one breath, lets the body drop weight downward
- [Single Breath Hum](https://protocolcards.com/#single-breath-hum) — one exhale with a hum, vagus nerve activation, completely covert

[…]

Example for a fictional person, created by the same generator that writes your human.md.

Built for People Who...

Already use AI daily and want it to understand you better

Know they need to "regulate" but are tired of generic advice

Have tried meditation apps but want something personalized

Don't want another platform, just want to upgrade what they have

If any of these resonate, your AI is ready for an upgrade.

Free to try. $15 to keep the full file.

Try the 15-minute interview for free and preview your opening section before paying anything.

Complete human.md

$15

One-time · No subscription

  • 15-minute voice or text interview with Atlas
  • Quick version (1,200 chars) for custom instructions
  • Full manual across 3 to 5 chosen focus areas
  • Matched practices linked to protocolcards.com cards
  • Prompts to explore plus instructions to keep it current
Start Free Interview →

Full refund if your file doesn't feel accurate.

Frequently Asked

It is a single plain-text markdown file that acts as an operating manual for AI about you. Inside, you get a 1,200-character Quick version for custom instructions, a full manual across 3 to 5 focus areas in your own words, matched practices on protocolcards.com, prompts to explore, and rules to keep the file updated.