The first AI wrapped
around the human psyche.
Every AI before COYL wrapped around external reality. We wrapped around the internal one.
Every AI wraps around something.
Look at the breakthroughs of the last three years. Each one picked a slice of external reality and built an interface into it. None of them touched the part of you that runs underneath conscious thought.
- Claude
- documents, code, the web
- ChatGPT
- language and knowledge
- MCP
- software systems
- Cursor
- codebases
Nobody has wrapped AI around the internal reality — the psyche. The pattern layer that runs underneath conscious thought. The part that opens the fridge before you’ve decided to.
COYL is the first AI that fires at the moment the psyche fires.
01
Doesn’t wait to be asked.
Every other AI sits in a prompt box and waits for you to start the conversation. COYL is the conversation that starts on its own — at the exact second your script is loading.
02
Learns the structural signature of your psyche.
Danger windows. Excuse taxonomy. Archetype. The shape of your autopilot is not the shape of someone else’s. COYL learns yours.
03
Wraps an intervention layer around that structure.
A precision interrupt, voice-matched to the user, calibrated to the moment. Not a daily affirmation. Not a weekly check-in. A 30-second call-out, positioned in the 3-second window.
04
Fires at the moment the psyche fires — not after.
Trackers tell you what happened. Coaches respond after you message them. COYL meets you in the moment before behavior happens, the only window where saying something useful changes the outcome.
The category name
Psyche AI. The behavioral interface layer.
The psyche has structure.
Autopilot behavior is not random. It is not chaos. It is a shape with measurable dimensions — the same way grammar has structure, the same way a heartbeat has structure. Four dimensions, all of them learnable.
Temporal
Fires in specific windows.
9–11pm. Monday morning. Post-lunch. The window where the script runs is the same window, week after week. Predictable in time.
Cue-based
Fires from specific triggers.
Stress. Boredom. Isolation. A location, a notification, a feeling. The pattern doesn’t fire arbitrarily — it fires from a finite set of cues.
Narrative
Runs through specific excuse scripts.
Eight classifiable categories. The sentence you say to yourself before you fold is not unique to the night. It is a script you have run before.
Recovery
Fails and rebounds in predictable sequences.
How you slip, how you spiral, how you come back — or don’t. The recovery curve is its own shape, measurable on its own.
Because the psyche has structure → it can be modeled → predicted → intercepted.
Why the data is the moat.
Every COYL interaction — every slip, interrupt, recovery, and excuse — builds a behavioral model of that individual’s psyche. Four properties make this dataset different from anything any other AI can collect.
Longitudinal
Months of data, not a session.
A chatbot sees you for one conversation. COYL sees the same window fire at 9:47pm twelve weeks in a row. Time is the signal no single-turn model can collect.
Pre-conscious
Captures behavior before rationalization.
A journal records what you tell yourself happened. COYL records what actually happened, in the second it happened, before your story about it gets written.
Ground-truth
Measures what people do, not what they report.
Self-report data is filtered through identity, mood, and memory. Behavioral signal is what the body actually did. The gap between the two is most of the failure mode of behavior change.
Population-comparable
Six archetypes cluster across millions of users.
Your 9 PM kitchen has a shape. So does your archetype’s. COYL learns the individual model and the population model simultaneously — the only way to give precision advice at scale.
This dataset is the moat. The app is how you build it.
The platform play.
MCP
Protocol between LLMs and software systems.
COYL Protocol
Protocol between LLMs and human behavioral reality.
The Behavioral Context Object any LLM can consume — no PII, only behavioral abstractions. A read API that turns any generic assistant into a psyche-aware advisor.
{
"archetype": "9PM_NEGOTIATOR",
"danger_window_active": true,
"window_confidence": 0.87,
"current_excuse_category": "DESERVER",
"self_trust_score": 74,
"risk_level": "HIGH"
}When Claude or GPT-4o has this object, they stop being generic assistants and become psyche-aware advisors. That’s the platform play. COYL owns the behavioral context layer no LLM can generate on its own.
The recurring anchor
AI for the moment before behavior happens.
Psyche AI is the category. COYL is the first product in it. If you build with LLMs, the protocol surface is where you start. If you live with your own autopilot, the audit is where you start.