All use cases

Upsell & Cross-sell

Next-best-product recommendation block

A customer mentions an upset stomach; the reply carries the answer — and a recommendation block with the reason why.

Next best thingBehavioral triggerChatIn-appHealth & TelehealthE-commerce & Retail

What it does

Customers reveal what they need in plain language all the time: "something for an upset stomach", "my skin gets dry in winter", "I'm starting to run again". Generic storefronts answer with a search box; most chat bots answer with text and lose the sale. This agent combines the live need with the customer's purchase history and renders a product recommendation block right inside its reply — a small set of fitting products, each with a one-line reason tied to what the customer just said. It is contextual cross-sell at the moment of need, never an interruption, and every outcome is written back to memory so recommendations get sharper over time.

How it works

  1. 1
    Trigger. mid-conversation, the customer asks a need-revealing question — a symptom, a goal, a situation — in the chat widget or in-app.
  2. 2
    Decision. the agent matches the stated need against the catalog and the customer's purchase history and memory, picks one to three genuinely fitting products, and attaches a one-line reason for each. If nothing fits well, it answers the question and recommends nothing.
  3. 3
    Action. the reply renders an inline recommendation block — product card, price, the reason — using the product's existing carousel and info-card components, so it looks native to the experience.
  4. 4
    Follow-through. taps, adds-to-cart, and dismissals are written back to memory; a dismissed category is not pushed again, and an accepted recommendation informs the next conversation.
Example
CustomerSomething for an upset stomach? Had a rough couple of days.
AgentSorry to hear that. For the next few days: a probiotic to settle your gut, and a rehydration mix since you'll want your electrolytes back up. The probiotic pairs with the digestive enzymes you bought in March. Want either added to your basket?
CustomerAdd the probiotic.
AgentIn your basket. Feel better soon — and if it hasn't settled in a few days, that's worth a chat with a clinician.

Configuration

How the agent is wired for this use case.

Triggermid-conversation, the customer asks a need-revealing question — a symptom, a goal, a situation — classified in-conversation on the chat or in-app surface.
Tools & actions
  • Product catalog · matches the stated need against attributes and stock to pick one to three fitting products
  • Messaging channel · renders the inline recommendation block (product card, price, one-line reason) using the existing carousel/info-card components
  • Cart API · adds an accepted recommendation to the customer's basket
  • Memory store · records taps, adds-to-cart, and dismissals so a dismissed category is not pushed again
Autonomymatching and the recommendation turn run unattended under judge gating; nothing is recommended when fit is weak, and regulated-product category rules govern what may surface.
Channelschat · in-app
Escalationwhere health signals warrant it, regulated-product category rules route the customer to a clinician instead of a recommendation.

What you need

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

Signals

need-revealing message classified in-conversation (symptom, goal, situational intent)

Data

product catalog with attributes and stock, purchase history, customer memory of preferences and past dismissals

Guardrails

recommend only on genuine fit — silence beats a forced offer; category rules for regulated products with escalation to a clinician where health signals warrant it; dismissal memory prevents repeat pushes; judge gating on the recommendation turn

Metrics it moves

  • attach-rateup, as need-matched products ride inside answers the customer asked for
  • conversion-rateup on recommendations versus interruptive banners, because the context earns the offer
  • aovup when the block surfaces the fitting companion alongside the answer

Related use cases

Upsell & Cross-sell

Second sale inside the support conversation

the same moment, framed from the support side

Upsell & Cross-sell

Complete-the-kit cross-sell

the cart-state sibling of this block

Upsell & Cross-sell

In-app proactive merchandising block

the agent-initiated version 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