All use cases

Engage / Lifecycle

Right-session-now meditation recommendation

The user opens the app stressed before a meeting and gets the one session that fits this moment, instead of a library to scroll.

Next best thingBehavioral triggerIn-appChatWellness (B2C apps)Subscription

What it does

A meditation app lives or dies on whether the user actually presses play, and the moment they open the app is the moment it can lose them. Faced with a wall of packs, sessions, and soundscapes, a user who came in stressed or short on time often browses, hesitates, and closes the app without meditating at all. This reads the moment the user is in — from a quick conversational check-in or from what their usage shows — and recommends the single best session or soundscape for right now, so the choice is made for them and they are listening within seconds. Less choice paralysis, more sessions started, a stickier daily habit.

How it works

  1. 1
    Trigger. the user opens the app, or types a short check-in into the assistant ("stressed before a meeting", "can't sleep", "ten minutes free"); the in-session context arrives — time of day, time available, stated mood, recent listening.
  2. 2
    Decision. the agent weighs that context against the content library and the user's history — what they have completed, what they abandoned, what length and voice they tend to finish — and selects the one session most likely to fit and to be finished, rather than returning a list. The recommendation clears a relevance and tone judge before it shows.
  3. 3
    Action. the app surfaces a single named session or soundscape with a one-tap play and a short reason ("five minutes, focused breathing, to settle before you walk in"), with a quiet "show me something else" fallback.
  4. 4
    Follow-through. the play, completion, and any "something else" tap write back to memory so the next recommendation is sharper; nothing clinical is asked or stored, and a check-in that surfaces distress routes to the app's support pathway rather than a meditation.
Example
Userstressed, big meeting in fifteen minutes
AgentThen let's keep it short. Try "Steady Before You Speak" — five minutes of slow breathing to settle your nerves, with time to spare before you walk in. Play it now?
Useryes please
AgentStarting it. If five minutes feels tight, there's a three-minute version — just say the word.

Configuration

How the agent is wired for this use case.

Triggeran app_opened event or a free-text check-in to the in-app assistant, which prompts a read of the current moment (time, availability, mood, recent activity).
Tools & actions
  • App backend · read the open event, session-time-of-day, and the user's stated mood or check-in text
  • Content library / catalog · match available sessions and soundscapes by theme, length, and intent to the moment
  • Customer memory · recall completed and abandoned sessions, preferred length and voice, and recurring contexts to personalize the pick
  • Recommendation service · score candidates and return the single best-fit session rather than a browse list
  • Messaging channel · render the one-tap recommendation with a short reason and a "something else" fallback in the in-app or chat surface
Autonomythe recommendation is composed and shown unattended under judge gating on relevance and tone; it suggests content only and takes no account or money action. This is a wellness content surface, not clinical: it never gives medical or mental-health advice.
Channelsin-app · chat
Escalationa check-in expressing distress, crisis, or self-harm language halts the recommendation and routes to the app's crisis/support pathway with a helpline and a human, 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

an app-open event and an optional free-text mood check-in, plus the in-session context to read at that moment — time of day, time available, recent listening.

Data

the content library with theme/length/intent tags, the user's listening history (completed, abandoned, preferred length and voice), and persistent memory of recurring contexts.

Guardrails

judge gating on every recommendation for relevance and tone; content-only scope with no medical or mental-health advice; crisis/self-harm detection that halts and routes to a helpline and a human; no clinical data stored; the client keeps owning the content surface.

Metrics it moves

  • click-through-rateup on the recommendation versus a browse screen, because one well-matched pick beats a list
  • session-lengthup, as users get into a fitting session fast instead of bouncing on choice paralysis
  • dau-mauup: an open that ends in a played session pulls the daily habit forward
  • churndown, since a habit that consistently delivers the right session in the moment is the habit users keep paying for

Related use cases

Engage / Lifecycle

Personalized conversation starters

the cross-vertical on-open opener this applies to meditation content selection

Engage / Lifecycle

Daily mood check-in companion

the recurring mood-log ritual whose answer can feed this session pick

Engage / Lifecycle

Proactive health-coach companion

the ongoing memory-led companion this in-session recommendation sits inside

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