All use cases

Retain / Save

Fasting-streak save with an earned skip

Your user is about to lose a months-long fasting streak over one off day — and they've already earned a skip that could save it, if anyone offers it in time.

Get aheadTime-based triggerBehavioral triggerPushIn-appWellness (B2C apps)Subscription

What it does

Fasting apps run on streaks, and a single missed window can wipe out months of progress in one stroke — which is exactly the loss-aversion moment that pushes a committed user to quit rather than start over from zero. Many of these apps already hand out a forgiveness mechanic, an earned skip or freeze the user banked for staying consistent, but it sits unused because nobody surfaces it at the moment it matters. The agent watches the fasting clock, and when a streak is about to break it offers to spend an earned skip to protect the run or right-sizes the fast to something the user can still finish. It turns a feature the user already earned into the thing that keeps them.

How it works

  1. 1
    Trigger. the fasting window is closing with no completed fast logged for the day, the streak is about to reset, and the user holds an earned skip or freeze token.
  2. 2
    Decision. the agent reads the user's memory — streak length, how they usually fast, whether today looks like a genuine off day or just a late start — and weighs offering the skip versus a shorter fast. Every unprompted message is judge-gated for tone, with quiet hours and frequency caps; the skip is presented as the user's own earned mechanic, not a discount or a guilt trip.
  3. 3
    Action. a short message on push or in-app — "use an earned skip to protect today's streak" with one-tap confirm, or a right-sized shorter window the user can still complete to keep the chain alive.
  4. 4
    Follow-through. on confirmation the token is spent and the streak is preserved; the choice writes back to memory so the next at-risk day is handled with context; the touch is cancelled the instant the user logs a fast on their own; impact is measured against a holdout.
Example
AgentYour fasting window closes soon and there's no fast logged yet — your months-long streak is on the line. You've earned a skip for staying this consistent. Want me to spend it to protect the streak today, or should I shorten today's window to something you can still finish?
UserUse the skip, today got away from me.
AgentDone — skip spent, streak safe. No reset, no starting over. You've still got a skip banked, so you're covered if another day like this comes up.

Configuration

How the agent is wired for this use case.

Triggera fasting_streak_at_risk event from the app backend — the fasting window is closing with no logged fast and the user holds an available skip/freeze token.
Tools & actions
  • Fasting app backend · read the fasting schedule, today's logged state, streak length, and skip/freeze-token balance
  • Token/forgiveness service · spend an earned skip on the user's confirmation to protect the streak
  • Persistent memory · recall the user's usual fasting pattern and prior at-risk days to choose between a skip and a right-sized fast
  • Quality judge · gate every unprompted message for supportive, non-coercive tone before send
  • Messaging channel · deliver the offer and one-tap confirm on push or in-app
Autonomythe at-risk detection and the offer run unattended under judge gating, quiet hours, and frequency caps; spending a token is taken only on the user's explicit confirmation, never silently. Wellness-coaching only — no medical, diagnostic, or clinical advice on fasting; under a wellness-safety policy class, any reply signalling disordered-eating patterns or a health concern is detect-and-escalate to a human.
Channelspush · in-app
Escalationa reply signalling disordered eating, faintness, or a health concern hands off to a human pathway with context; a user who asks to stop is honored immediately.

What you need

The inputs this use case runs on. Your channels stay yours; the agent supplies the judgment.

Signals

fasting-window-closing events, today's logged-fast state, streak length, skip/freeze-token balance, self-logged fasts, user replies

Data

the user's fasting schedule and usual pattern, streak history, token balance, channel preference and consent, quiet hours, persistent cross-session memory

Guardrails

wellness-coaching only with no clinical fasting advice; disordered-eating and health-concern detection with human escalation; token spent only on explicit confirmation; judge gating on every unprompted message; consent, quiet hours, and frequency caps; stale-touch cancellation on self-logged fast; holdout measurement; the client keeps owning the send channels

Metrics it moves

  • churndown, because a single off day no longer wipes a months-long streak and tips a committed user into quitting
  • adherenceup, as the forgiveness mechanic keeps the fasting routine intact through an unavoidable miss
  • dau-mauup, since a protected streak keeps the daily return habit alive, measured against the holdout

Related use cases

Retain / Save

Habit-streak break recovery

the post-break recovery for when no token was spent and the chain reset

Retain / Save

Logging-streak rescue save

the pre-break logging-streak save this shares a loss-aversion trigger with

Retain / Save

Adherence guardian for chronic-condition programs

the risk-window retention pattern this applies to a fasting routine

See it on your own customer journey

Bring one drop-off, one churn cliff, or one silent segment. We will show you what a proactive agent with memory and judgment does with it.

Book a demo