Manifestation with a scoreboard.

The inversion

The whole intention-setting category sells belief and avoids measurement. Latent does the opposite.

It turns an unfalsifiable question — "does this work?" — into a measurable one: which of these habits improves my hit rate? Latent doesn't know whether intention-setting works. Neither do we — that's exactly why it's worth measuring, honestly, against your own record.

How an entry works

1. Wish

   What you want, written as it feels — no editing yourself into a spreadsheet yet.
2. Make it real

   A checkable criterion and a deadline. Gentle, advisory hints nudge the one question that matters: could future-you answer this yes or no?
3. Honest prior

   On a 1–99 scale: how likely is this anyway? An honest 20% is what makes a later yes mean something.
4. Practices

   Toggle any of four evidence-backed habits — next action, if-then plan, mental rehearsal, gratitude — each shown with its research citation.

Later you resolve each intention as hit, partial, miss, or void. Onboarding seeds a few fast-resolving intentions, so your first result lands within a couple of days.

Why raw priors and frozen practices

Here you want the outcome, so a low honest prior is normal and correct — Latent keeps priors raw, with no negation flip. Your practices are tagged before you know the result and frozen after you resolve, so hindsight can't reach back and credit the habit that merely happened to coincide with a win. That freeze is what makes the "what actually works for you" view trustworthy.

How the score is made

Lift is the headline: your hit rate minus your own average prior, in points.

It's the stat no belief-app will show you. Alongside it: a Brier score, a calibration read, and the money view — your per-practice hit rate with a habit versus without it, shown as a ratio, revealed only once both sides have enough data, and always labeled correlation, not proof.

The AI — checkable, not magical

A free, on-device writing assist turns a vague wish into a concrete, scoreable criterion. It suggests checkable phrasing and nothing more — it never promises an outcome, never invents a number, and never gives medical or financial advice. A hard honesty filter, not the prompt, is what guarantees that. It's free with a supported device because it's a writing aid, not a premium prophecy — and it's part of logging, which is always free.

Features

• Guided captureFour steps — wish, checkable criterion and deadline, honest 1–99 prior, practices.
• Four practicesNext action, if-then plan, mental rehearsal, gratitude — each with its citation, frozen at resolve.
• Honest outcomesResolve as hit, partial, miss, or void — voids counted, never scored.
• LiftYour hit rate against your own average prior, in points.
• Brier & calibrationA calibration read and a widest-gap verdict on where you're over-hopeful.
• Practice correlationHit rate with a habit vs without it, as a ratio — shown only with enough data, labeled correlation.
• Writing assistFree, on-device suggestions that make a wish checkable — never a number, never a promise.
• Seeded startA few fast-resolving intentions so your first result lands within a couple of days.

Privacy

Latent runs entirely on your device — no accounts, no analytics, no servers. The writing assist runs on-device too: nothing you write leaves your phone. Optional Pro iCloud sync syncs through your own private iCloud.

Status