Every morning, a mission shaped for you: this flow walks from your onboarding answers to the quest you see in the app — with a spark of creativity and guardrails that keep you safe.
Pipeline
Profile & memory
During onboarding you share how you like to play in real life — explorer vs homebody, cautious vs bold — plus a light personality footprint that steers recommendations without showing clinical jargon in the app.
Each completed, accepted, or declined quest leaves a gentle trace. Recent choices feed the engine’s memory — not to judge you, but to understand how you actually move through the world.
The system infers tendencies from the kinds of quests you’ve lived. It’s a soft read on behavior, not a box you’re locked into.
We measure how closely what you say about yourself aligns with what recent quests suggest. That signal tunes tone and intensity — always supportive, never guilt-tripping.
Context & constraints
Calibration (near comfort), expansion (small stretch), rupture (bigger challenge): the phase targets the kind of nudge that fits where you are right now.
Weather, city, and — if you enable location — whether an outdoor outing makes sense today. When it fits the mission, we may suggest a public place on the map.
The engine still scores quest families (affinity, phase, freshness, preferences), then picks today’s family, a target intensity, and an ideal duration — using weather, place, and history. The taxonomy is inspiration only: a few short examples may appear in the brief, not a template to copy.
Writing & safeguards
A language model invents the title, hook, and mission from those constraints and today’s context. It’s no longer picking from a list of archetypes: the quest is written for you in a warm narrator voice, with safety, clarity, no medical or dangerous advice, and respect for your language.
Automatic rules validate the mission: one actionable sentence, no forbidden content, outdoor place consistency when needed, city mention when GPS is on. A taxonomy id is then attached behind the scenes for the engine’s memory — the copy stays fully creative. If validation fails, we retry or fall back to a safe taxonomy entry.
One quest is stored per calendar day: reopening the app the same day shows the same mission — no spam regenerations. Tomorrow the journey runs again with fresh context.
Nothing is pure randomness: the engine blends your profile, your history, and the day ahead, then sets today’s trajectory (quest family, intensity, duration). AI writes the words from there; Questia enforces safety, clarity, and respect for your pace.
You won’t see clinical labels: copy talks about tendencies and rhythm, not diagnosis. Data is used to personalize the experience, as described in our privacy policy.