Engage / Lifecycle
Cycle-phase prediction push
The model already knows her period is two days out — instead of a generic banner, she gets a push that speaks to the phase she's about to enter and opens straight into a read worth her time.
What it does
A cycle tracker quietly predicts the next phase change every day, and most apps spend that prediction on a static pop-up — "your period is coming" — that users mute within a week. The prediction itself is one of the strongest engagement signals the product has, and it is wasted on a template. The agent turns each predicted phase entry (period-soon, PMS window, ovulation) into a personalized, content-linked push, written for what she is about to feel and tappable into her advisor or the matching content. It is the conversational upgrade of the static cycle notification, made affordable to send at scale once a small fine-tuned model brings the per-message cost down.
How it works
- 1Trigger. the prediction engine emits a
phase_transition_predictedevent a set number of days before a forecast phase change, with the predicted phase and date. - 2Decision. the agent reads her recent logging, remembered phase patterns, and consent and quiet-hours state, then drafts a short message tied to the specific upcoming phase; a quality judge scores relevance and timing, and a low-value or mistimed send is dropped rather than fired.
- 3Action. a personalized push goes out through the client's own provider, deep-linking into a phase-matched content read or the advisor thread; the same surface can render in-app for users who have push switched off.
- 4Follow-through. if she opens the app on her own or the predicted date passes before the send fires, the queued push is cancelled; opens and taps write back to analytics, and lift is measured against the templated-notification baseline.
Configuration
How the agent is wired for this use case.
phase_transition_predicted event from the prediction engine a set number of days before a forecast phase change, carrying the predicted phase and date.- Prediction engine · read the forecast phase, the predicted date, and confidence.
- Memory store · load recent logging, remembered phase-by-phase patterns, and prior reactions.
- Knowledge base · pull the phase-matched educational content the push links to.
- Messaging channel · send the personalized push through the client's own provider and deep-link into content or advisor chat.
- In-app surface · render the same prompt for users who have push disabled.
- Product analytics · receive opens and taps back and measure lift against the templated baseline.
What you need
The inputs this use case runs on. Your channels stay yours; the agent supplies the judgment.
Signals
a phase-transition prediction event (predicted phase, date, confidence) from the cycle-prediction engine.
Data
per-user logging history, remembered phase patterns, phase-matched content library, notification consent and device tokens.
Guardrails
judge review of every push with "don't send" available, frequency caps and quiet hours, instant opt-out, cancellation of stale sends, trends-and-education framing with no diagnosis or medication.
Metrics it moves
- dau-mauup: a prediction that used to sit silent now opens a session at the moment she's most curious about her own body.
- open-rateup: a push tied to the phase she's about to feel beats a generic "your period is coming" template.
- click-through-rateup: the deep link lands on phase-matched content or the advisor, not a static screen.
- opt-out-ratedown: relevance plus the judge's silence option replace notification fatigue.
Related use cases
Phase-aware proactive hero card on a cycle log
the log-triggered in-app sibling of this prediction-triggered push
Proactive bubbles and personalized push
the generic recent-activity surface this femtech phase-timing push specializes
Hormone-result interpretation grounded in trend history
where the advisor thread the push opens into goes deeper
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