All use cases

Onboard / Activate

New-player activation and first-session companion

Most players who quit a game never quit because they hated it — they quit before they finished the tutorial, and the agent is the in-game guide that gets them to the first moment it gets fun.

Next best thingGet aheadBehavioral triggerTime-based triggerSystem triggerIn-appPushChatGaming

What it does

A new player decides whether to stay in the first session or two, and the decision is made on a handful of milestones: finishing the tutorial, getting through the first real loop, making a first top-up, finding other people to play with. Players who hit those early stick; players who stall on them churn silently and never come back. The companion lives inside the game during those first sessions: it nudges the next activation step at the right beat, unblocks a player stuck in onboarding, points first-timers toward a guild or a live event, and answers the "how do I…" question in-context — so more new players reach the sticky first-value moment instead of bouncing at the loading screen.

How it works

  1. 1
    Trigger. in-game behavioral, time, and system signals — tutorial_step_skipped, a core-loop never attempted, a first-session timer elapsing without progress, an onboarding stall flagged by the game client.
  2. 2
    Decision. the agent weighs where this player is against the activation path that predicts retained players, recalls what they have already done and asked this session, and picks one next step — not a wall of tips. Every unprompted message passes a quality judge before it is shown.
  3. 3
    Action. an in-app prompt or push that points at the exact next move — finish the tutorial step, try the core loop, claim the starter reward, join a beginner guild or the running event — with any question answered in the same conversation, never interrupting active play.
  4. 4
    Follow-through. if the player completes the step themselves, the queued nudge is cancelled before it sends; activation milestones write back to the game's analytics; lift is measured against a holdout of new players, not assumed.
Example
AgentNice — you cleared the tutorial. The fastest way to see what makes this fun is the first raid; it takes about five minutes and you'll unlock your starter loadout. Want me to drop you into one?
Customeris it hard? I don't want to get wrecked
AgentIt's the beginner version, matched to your level — low stakes, good loot. I can also pop you into a starter guild first so you're not going in solo. Which sounds better?

Configuration

How the agent is wired for this use case.

Triggeran in-game activation signal (tutorial_step_skipped, core_loop_unattempted, first-session stall timer, feature_unused) emitted by the game client or live-ops platform.
Tools & actions
  • Game telemetry / events · read tutorial progress, core-loop attempts, session activity, and first-deposit state to place the player on the activation path.
  • Player memory · recall this player's progress, questions, and preferences within and across sessions to pick one relevant next step.
  • Live-ops / rewards system · surface and grant the starter reward or first-step incentive, and point to the active event.
  • Social / guild system · match a new player into a beginner guild or group so the social hook lands early.
  • In-app messaging and push · deliver the judge-gated nudge or deep-link without interrupting active play, and cancel it if the step is self-completed.
Autonomystall detection, next-step selection, deep-linking, and starter-reward grants run unattended under judge gating, frequency caps, and quiet-hours rules, with lift measured against a new-player holdout. Anything touching real-money purchase (a first top-up) only guides the player into the platform's own purchase flow; the agent never takes payment itself.
Channelsin-app · push · chat
Escalationa payment or account issue, a suspected underage or responsible-play concern, or a question the agent cannot resolve hands off to player support or the relevant team with session context attached.

What you need

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

Signals

game events (tutorial_step_skipped, core_loop_unattempted, session activity, feature_unused) plus first-session and onboarding stall timers.

Data

the activation path itself (which early milestones predict retained players), the player's progress and preferences this session, the rewards and event catalog, guild/social structure, notification preferences and consent.

Guardrails

frequency caps and quiet hours so the companion never interrupts active play; judge review of every unprompted message; responsible-play and underage protections; real-money steps go to the platform's own purchase flow, never an agent-taken payment; holdout measurement so claimed activation lift is proven.

Metrics it moves

  • activation-rateup: by walking each new player through the early milestones that predict retention instead of leaving them to find the fun alone.
  • feature-adoptionup: deep-linked nudges put the untried core loop, event, or social feature in front of the right player at the right beat.
  • churndown: players who reach first value in the opening sessions are the players who keep coming back.

Related use cases

Onboard / Activate

In-app activation companion

the cross-industry parent this sharpens for the first-session gaming case

Prevent / Care

Preventive outreach on gaming/account issues

the proactive save when a new player hits a top-up or session problem

Engage / Lifecycle

Gaming lifecycle and event engagement

the ongoing engagement that takes over once the player is activated

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