All use cases

Upsell & Cross-sell

In-app proactive merchandising block

The shopper searches, browses, lingers — and the shelf quietly rearranges itself around them, before anyone opens a chat.

Get aheadNext best thingBehavioral triggerTime-based triggerIn-appPushE-commerce & Retail

What it does

Most personalization waits for the customer to start a conversation, so the shoppers who never open the chat widget never see it. This is an always-on surface inside the app or site that reacts to live behavior — the search query just typed, the category just browsed, dwell on a product, what's sitting in the cart — and rotates in the recommendation or promo that fits this moment. A returning customer gets a memory-based offer at app open; a shopper who just toured a category sees a voucher for the adjacent product they'd naturally buy next. A never-empty fallback means the block always shows something sensible, even for a first-time visitor with no history.

How it works

  1. 1
    Trigger. behavioral events stream from the app or site — search queries, category and product views, dwell time, cart changes, app opens.
  2. 2
    Decision. the agent scores candidate offers against the live session and the customer's memory (past purchases, offers already shown, declines), with rotation logic and a never-empty fallback. A quality judge gates anything carrying a discount, and offers made stale by newer behavior — the item is already in the cart, the voucher would now double-discount — are pulled before they render.
  3. 3
    Action. the block re-renders in place inside the host product UI; for app-open offers, a push notification can carry the same offer to bring the customer back in.
  4. 4
    Follow-through. a tap opens the product page or a pre-filled cart, impressions, clicks, and conversions are written back to memory and analytics, and a holdout slice of traffic measures the lift the block actually causes.
Example
Agent (in-app block)You've been comparing trail runners — here's 20% off the pair you keep coming back to, valid today.
Customer (taps through)Does that apply to the waterproof version too?
AgentIt does — the voucher covers the whole trail line. I've applied it to your cart, so it's there whenever you're ready to check out.

Configuration

How the agent is wired for this use case.

Triggera behavioral event stream from the host app or site — search queries, category and product views, dwell time, cart changes, app opens.
Tools & actions
  • Host product UI slot (embed or SDK) · re-renders the merchandising block in place with the chosen offer
  • Product catalog · reads live stock and pricing to score and validate candidate offers
  • Promotions engine · resolves voucher rules and discount eligibility, applies the voucher to the cart
  • Messaging channel · sends an app-open push carrying the same offer to bring the customer back
  • Analytics store · writes impressions, clicks, and conversions back for memory and holdout measurement
Autonomyoffer selection, rotation, and re-render run unattended, judge-gated, with the quality judge gating anything carrying a discount and stale offers pulled before render; discount-budget and voucher-eligibility limits are enforced before any money-moving discount renders.
Channelsin-app · push
Escalationno human handoff in-flow; a holdout slice of traffic is held back so reported lift stays causal, and offers made stale by newer behavior are cancelled before they render.

What you need

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

Signals

search queries, category and product view events, dwell time, cart events, app-open events

Data

product catalog with live stock and pricing, promo and voucher rules, customer memory (purchase history, offers shown and declined), current session context

Guardrails

discount budget caps and voucher eligibility enforced before render; judge gating on every promo; rotation and frequency caps so the block recommends rather than nags; never-empty fallback content; holdout measurement so reported lift is causal

Metrics it moves

  • conversion-rateup, because the offer appears at the moment of demonstrated intent instead of waiting for a chat
  • aovup through adjacent-product vouchers that extend the basket the shopper already started
  • attach-rateup on companion products surfaced right after the category browse that signals the need

Related use cases

Upsell & Cross-sell

Next-best-product recommendation block

the same surface inside a conversation

Convert

In-session hesitation rescue

the chat takeover when browsing turns into stalling

Engage / Lifecycle

Proactive bubbles and personalized push

the engagement-lane sibling of this surface

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