Know yourself. With evidence this time.
Selph turns a conversation into a working model of your psychology — then tests it with small experiments and learns from what actually happens.
Private, invite-only beta, free for invited users. Your graph stays on your account, yours to export or delete anytime.
How it works
One conversation. A graph that grows as you speak.
Pick a starting place. Selph asks one question at a time and quietly maps what you say into your private graph.
Just a taste — nothing is saved, and this runs entirely in your browser.
Patterns become something you can test.
After a few sessions, Selph notices something stable enough to name. It ties the insight to a behavior or outcome in your graph and ranks it by how much it would matter if true.
You're acting as if going to bed will elicit next-day exhaustion. That's a claim we can falsify — by getting in bed on time and checking how tomorrow actually feels.
anchored to behavior · going to bed → elicits → outcome · next-day exhaustion · supported by 3 sessions
Lights out by 11
A 1-week experiment on whether going to bed elicits next-day exhaustion
- If
- it's past 10:30 on a weeknight,
- Then
- I'll put the phone in another room and get in bed within 15 minutes.
- Predicts
- Tomorrow won't feel worse — or the exhaustion spiral won't show up.
- Disconfirms
- If I'm clearly exhausted twice after early nights, the "elicits exhaustion" edge still stands for now.
A hypothesis earns its place by surviving real life.
Selph designs the smallest experiment that could change its mind — with a clear prediction and an honest way to disconfirm it.
Your experiment follows you into the day.
Selph nudges you in the moments that matter — not on a schedule. The phone gets the smallest possible reminder; the heavy lifting stays in the conversation.
- Tied to one experiment, not your whole life
- Snooze, skip, or "didn't apply" — all first-class
- Quiet hours and contextual cool-downs by default
9:41
Tuesday, May 12
Wind-down window?
Phone down by 10:30. Test whether going to bed actually elicits tomorrow's exhaustion.
What you learned changes the graph.
This is one full loop, start to finish: a hunch about restless evenings became a testable hypothesis, a week's experiment produced evidence, and the graph updated — the “doomscrolling” edge faded, “going to bed” earned a rested-next-day outcome. An illustration of the loop, not a real person's data.
Before
triggersdrivesinhibitspreventselicitsfrustratesAfter
triggersdrivesinhibitspreventselicitsfrustratesleads topreservesWho Selph is for.
For you if
- You want to understand your own patterns, not just track habits.
- You'd rather test a belief about yourself than be told a personality type.
- You're willing to be honest with a tool that reflects back what it sees.
Not for you if
- You're looking for therapy, diagnosis, or crisis support.
- You want quick affirmations more than an honest mirror.
- You don't want your inner life written down anywhere, even privately.
Selph's values are spelled out in full — read the Manifesto.
You're handing Selph something intimate. Here's the deal.
Private by default
Your graph lives on your account. It isn't public, and it isn't shared.
We don't train on your data
Your conversations and graph are never used to train Selph's models without your consent.
Yours to export or delete
Ask for everything we hold, or ask us to delete it. It's your record of yourself.
Read the full Privacy page — plain English, including what's still shipping for the closed beta.
Start your graph.
Session one is a single conversation — a few nodes and your first named pattern. The rest of the loop unlocks as your graph has more to say.
Invite-only beta, free for invited users · Private — see Privacy for export and delete · iOS app in private beta