All use cases

Engage / Lifecycle

Daily behavior-change micro-lesson coach

Each day a two-minute read lands that speaks to what your own week actually looked like — and ends with one small thing to try, so the app teaches you why your habits happen, not just what you ate.

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

What it does

Tracking what you eat tells you the numbers but never changes the behaviour behind them, and a generic article library goes unread because nothing in it is about this user. That gap is why nutrition programs stall after the novelty fades and the subscription lapses. The agent runs a daily behaviour-change lesson grounded in psychology, but it picks and frames each lesson from the user's recent logs — connecting a pattern they actually showed (stress before snacking, skipped breakfasts, weekend slips) to a short, readable lesson and one tiny action to try today. It is the daily curriculum engine that defines the behaviour-change category and keeps a multi-week program feeling alive rather than abandoned.

How it works

  1. 1
    Trigger. the daily lesson window opens, and the agent looks for a behavioural pattern in the user's recent logs (mood-before-eating, timing, repeated food choices) worth teaching into.
  2. 2
    Decision. it selects the next lesson from the curriculum that fits both the user's program stage and the pattern in their data, avoiding repeats and respecting where they are in the multi-week arc; every unprompted send passes a quality judge with quiet hours and frequency caps.
  3. 3
    Action. it delivers a short, personal lesson on the user's preferred channel — naming the pattern, giving the two-minute read, and proposing one small action — tappable into a fuller in-app lesson or a chat.
  4. 4
    Follow-through. lesson completion and the user's response write back to memory so tomorrow's lesson builds on it rather than repeating; a queued lesson is cancelled or reselected if the user's logs shift before it lands, and engagement is measured against a holdout so the retention lift claimed is the lift measured.
Example
AgentTwo-minute read for today. Looking back at your week, you logged "stressed" right before three of your evening snacks — that's emotional eating, and it's a pattern, not a willpower problem. The lesson: name the feeling before you reach for food. Tonight's tiny action — when the urge hits, pause and text yourself the word for what you're feeling first. Tap to read why it works.
Userthat's actually really me lol
AgentMost people, honestly — and naming it is the part that loosens its grip. Try the pause tonight and tell me tomorrow how it went. We'll build on it.

Configuration

How the agent is wired for this use case.

Triggerthe daily lesson_window timer combined with a behavioural pattern detected in the user's recent logs (mood, timing, food-choice signals).
Tools & actions
  • App backend · read recent food, mood, and timing logs to find the pattern the day's lesson should address
  • Knowledge base · select the next psychology-grounded lesson from the curriculum that fits the user's stage and pattern
  • Persistent memory · track which lessons were delivered and completed and what the user responded to, and write today's outcome back
  • Messaging channel · deliver the short personalised lesson plus its one tiny action on the user's preferred surface
  • App backend · deep-link into the full in-app lesson and log completion
Autonomypattern detection, lesson selection, and the daily send run unattended under judge gating, quiet hours, and frequency caps. Behaviour-change and nutrition-psychology education only — no medical, clinical, or eating-disorder treatment; under a wellness-safety policy class, any log or reply signalling disordered eating or distress is detect-and-escalate to a human, never coached by the agent.
Channelspush · in-app · chat
Escalationlogs or replies indicating disordered-eating patterns, a mental-health concern, or distress hand off to a human or the program's appropriate pathway with context rather than being met with another lesson; a user who asks to pause the daily lessons is honoured immediately.

What you need

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

Signals

the daily lesson-window timer, recent food/mood/timing log events, lesson-completion events, and user replies

Data

the psychology-grounded lesson curriculum, the user's program stage and lesson history, recent logs and patterns, goals, channel and notification preference, consent, quiet hours, persistent memory

Guardrails

behaviour-change and nutrition-psychology education only — no clinical or eating-disorder treatment, with a stated boundary and human escalation on a disordered-eating or distress signal; judge gating on every daily lesson; frequency caps and quiet hours; stale-lesson cancellation when logs shift before it lands; holdout measurement; the client keeps owning the send channels

Metrics it moves

  • dau-mauup: a lesson that is visibly about the user's own week earns the daily open a generic article library never would
  • adherenceup: connecting behaviour to a tiny next action turns passive logging into habit change the program depends on
  • ltvup: a curriculum that keeps unfolding sustains the multi-week program past the novelty drop-off, measured against the holdout

Related use cases

Engage / Lifecycle

Photo food-logging assist and confirm

the effortless logging that produces the patterns these lessons read from

Engage / Lifecycle

Proactive health-coach companion

the reactive coaching companion this differs from by being a scheduled curriculum engine

Retain / Save

Weight-plateau coaching intervention

the save that fires when behaviour-change alone stops moving the number

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