All use cases

Engage / Lifecycle

Personalized journaling prompt of the day

Instead of a blank page, the user opens to "you mentioned work stress twice this week — what's one thing in your control tomorrow?" and starts writing.

Get aheadNext best thingTime-based triggerBehavioral triggerPushIn-appChatWellness (B2C apps)Subscription

What it does

The blank page is what kills a journaling habit. A user who downloaded the app to reflect opens it, sees an empty box and a blinking cursor, has nothing to say, and closes it — and a few empty days later the habit is gone and so is the subscription. This serves a single prompt of the day drawn from the user's own recent entries, so the page already has a thread to pull. Because the prompt recalls what they have actually been writing about, it lands as continuity rather than a generic question, lowers the friction that ends journaling habits, and raises how often they come back and write.

How it works

  1. 1
    Trigger. the user's usual journaling window arrives, or they open the journal to an empty entry; a time-and-behavioral signal fires the prompt of the day.
  2. 2
    Decision. the agent reads the user's recent entries from memory, finds a recurring theme or an open thread worth returning to, and composes one prompt that builds on it — weighing tone, recency, and whether a lighter or deeper prompt fits today. The prompt clears a relevance and tone judge before it shows, with a neutral evergreen prompt as the fallback when there is nothing specific worth surfacing.
  3. 3
    Action. the app presents the single prompt with a one-tap start to write, plus a "give me another" option, on the channel the user engages with.
  4. 4
    Follow-through. the new entry writes back to memory so tomorrow's prompt stays current; the pending prompt is recomputed or dropped if it goes stale before they return; an entry that surfaces distress or self-harm language routes to the app's support pathway with a helpline and a human; entry frequency is measured against a holdout.
Example
AgentToday's prompt: you wrote about feeling stretched at work twice this week. What's one thing tomorrow that's actually in your control — and one you can let go of?
Userhonestly I keep saying yes to everything
AgentThat's worth writing down. Try finishing the thought: "The next thing I'll say no to is…" — three lines is plenty. I'll pick up the thread tomorrow.

Configuration

How the agent is wired for this use case.

Triggera journal_window_due time signal or an empty_entry_opened behavioral event, which prompts a read of the user's recent entries.
Tools & actions
  • Journal store · read recent entries to find a recurring theme or an open thread to build the prompt on
  • Customer memory · recall the user's themes, tone, prompts they engaged with, and quiet hours to personalize and pace
  • Content/prompt library · supply themed and evergreen prompt templates the agent personalizes or falls back to
  • Messaging channel · deliver the single prompt with a one-tap start and a "give me another" option on push, in-app, or chat
  • App backend · write the new entry back and update the prompt state so the next day's prompt stays current
Autonomythe daily prompt is composed and shown unattended under judge gating on relevance and tone, with quiet hours and frequency caps. It offers reflective, non-clinical prompts only — no diagnosis, no medical or mental-health advice.
Channelspush · in-app · chat
Escalationan entry expressing crisis, self-harm, or acute distress halts the prompt flow and routes to the app's crisis pathway with a helpline and a human, and the loop is closed back to safety, per the mental-wellness safety policy.

What you need

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

Signals

a journaling-window time signal or an empty-entry open event, plus the user's recent entries to draw on.

Data

the user's prior journal entries and themes, prompts they have engaged with, channel preference and consent, quiet hours, and persistent cross-session memory.

Guardrails

reflective non-clinical prompts only — no diagnosis or medical advice; crisis/self-harm detection that halts the flow and routes to a helpline and a human with loop closure; judge gating on every prompt for relevance and tone; a neutral evergreen fallback when nothing specific is worth surfacing; recompute or drop a stale prompt; consent, quiet hours, and frequency caps; holdout measurement; the client keeps owning the send channels.

Metrics it moves

  • adherenceup on journaling frequency, because a prompt that continues the user's own thread beats a blank page
  • session-lengthup, as a prompt with something to pull on opens into a real entry rather than a bounce
  • dau-mauup: a daily prompt worth answering pulls users back into the writing habit
  • churndown, since the reflection habit that sticks is the subscription that renews

Related use cases

Engage / Lifecycle

Personalized conversation starters

the cold-open opener this differs from by recalling the user's own prior entries rather than current live context

Engage / Lifecycle

Daily mood check-in companion

the sibling reflective ritual, a one-tap mood log versus prompt-led writing

Prevent / Care

Crisis-language detection and safe handoff

the safety rail this prompt escalates into when an entry signals distress

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