AI Agent for eCommerce Upselling and Cross-Selling (2026)

Monday at checkout: a camera body sits in the basket. The drawer offers a lighting kit that does not mount. Collaborative filters call that an attach. Your returns pile calls it something else.

The useful upsell is not “people also bought.” It is the next SKU that still matches the cart: fits, in stock, not already owned, not a restricted add-on, not a price rewrite.

Baymard puts average cart abandonment at 70.22% across 50 studies (updated Sep 22, 2025). That is checkout leak context. It is not a promise that add-ons will save it.

This post is cart-aware ranking. Discovery-from-intent lives in AI product recommendation agents. Catalog completeness lives in catalog management. Do not collapse the three.

This is post 15 in the 15-automations map. We have no published store results that say attach rate or average order value moved.

The job. Suggest the next SKU that fits the cart — relationships, stock, history — not co-occurrence alone.

This week. One accessory family with a real fits graph. Lookups only. Present for confirm. No silent cart-add. No price cut.

A person still signs. Published bundles, margin floors as filters, discounts (separate path), and silent cart mutation later.

Skip it when a three-line merch rule already maps body → lenses correctly, compatibility is missing from the catalog, or leadership wants quiet attach for margin.

Our take: next-best is contextual. Without a compatibility graph, ship a static merch rule and skip the agent.

Copy the matrix — Open upsell-cross-sell-decision-matrix.md. Fill Owner and Approval. Co-occurrence is secondary. Ship gates: monday-checklist.md.

Why “people also bought” fails at the cart

Checkout and cart drawers show add-ons. Most are a merch pin or a frequently-bought-together table. They do not know the shopper already bought the spare battery. They do not know the lens mount. They will still “upsell” a thinner-margin hero if the model is told to maximize average order value.

A bad attach is a return and a support ticket. A good attach is guided selling at the moment of intent. Who breaks without compatibility fields: anyone selling accessories, parts, or sized goods.

How stores handle this today

Shopify apps, “complete the look” pixels, rules in the order management system (OMS), sales associates on the phone. Rules win when the graph is small and correct (this SKU → these three SKUs). Agents win when the cart is messy (two categories, a constraint in chat, a size already chosen). Hybrid: rules for the obvious edge, agent for the rest — agent vs workflow.

What runs this week (lookups and drafts)

• Upsell: same family, higher capacity / warranty, still constraint-safe, cited.
• Cross-sell: compatible accessory or consumable, in stock, not already owned.
• Bundle candidate: structured set for a merchandiser to publish.

Margin and live promos are inputs, not the objective function. A high-margin unrelated SKU must not win.

What a person still owns

• Compatibility and restricted flags.
• Inventory and price lookup (no writes).
• Promo engine eligibility for “this add-on is on a deal” — display only.
• Cart mutation policy (shopper confirm).

No price writes. If they want a cheaper add-on, that is personalized offers plus a person approving, not setPrice.

Merch owns the relationship graph and published bundles. Finance owns margin floors as filters, not as a ranking god. CX handles “I already have this” when history tools are wrong. Nobody should need a human to suppress an out-of-stock add-on if inventory is fresh.

flowchart TD
  cart[Current cart]
  agent[Upsell agent]
  tools[Relationships inventory history margin promo]
  next[Next-best SKU]
  present[Present for confirm]
  skip[Skip or escalate]
  cart --> agent
  agent --> tools
  tools --> next
  next -->|compatible in stock| present
  next -->|already owned restricted unknown| skip

Lookups you attach this week

Week-one allow-list:

Writes week one: none. Later addToCart with Cedar and confirm. Payment capture out of tools. Memory: this cart. Do not store a “always upsell” preference that overrides compatibility.

There is no native Shopify connector.

Ship one accessory family first

One accessory family with a real fits-graph. Goldens: compatible add-on, already-owned skip, out-of-stock skip, margin-only SKU must not appear, price-cut request must not write.

If a three-line merch rule already maps body → lenses correctly, leave it. Agent the messy carts.

For your technical lead

On June 17, 2026, Amazon Bedrock AgentCore Harness reached general availability (What’s New). Agents Classic is in maintenance for new customers after July 30, 2026.

Bedrock is the model layer, not the host. Harness or Runtime hosts the loop. Gateway + Cedar on cart writes. Browser and Code Interpreter off. Next.js is storefront or human-in-the-loop, not the runtime. Strands does not provide Gateway, Identity, Policy, or microVMs.

Share catalog tools with the recommendation agent. Separate prompts so a discovery session cannot silently attach. Supervisor + hop cap if those specialists coexist (store-agents).

First-party signals we reuse (not eCommerce client outcomes) — Gateway server-side tools cut median tool round-trip ~180 ms → ~95 ms on a B2B CRM assistant (12 tools, ~8k turns/day) — Gateway post. Platform TCO silhouette: support-style AgentCore at 50K sessions/mo ~$791/mo platform + model (decision guide). Model widget volume on the AgentCore pricing calculator.

Gateway ~95 ms median on the CRM canary is not your cart API’s p95.

What broke — Cart-drawer prototype ranked add-ons from frequently bought together plus a “maximize margin” instruction. A camera body received a high-margin lighting kit that used a different mount. Detection: golden “body in cart → compatible lens” failed; traces showed getFrequentlyBoughtTogether and getMargin, no getCompatibility. Recovery: compatibility required; margin tie-break only; fail evals if co-occurrence is the only evidence. Lesson: attach rate without fit is a return rate you have not measured yet — and we are not inventing yours.

What to do this week

1. Pick one family. Confirm relationship fields exist in the PIM. If not, catalog first.
2. Copy upsell-cross-sell-decision-matrix.md. Owner and Approval on auto-attach (default: no).
3. Tools: cart, relationships, inventory, history. No setPrice.
4. Harness; Policy LOG_ONLY; default-deny writes; Browser off.
5. Five goldens above. Alarm on unexpected tool mix.
6. Presentment only — no silent cart mutation.
7. Price on the AgentCore pricing calculator. Run monday-checklist.md.

Need Gateway on cart + PIM before an attach widget goes live? Contact us. Related: Amazon Bedrock consulting, Generative AI on AWS, AWS for retail / eCommerce.

What this post doesn’t cover

• Session-intent recommendations — recommendation agent.
• Discount issuance — personalized offers.
• Homepage merchandising — dynamic merchandising.
• Measured attach rate, average order value, or conversion from a client engagement.
• Payment capture / x402.
• A native Shopify AgentCore connector.

FAQ

When should we NOT run an AI upsell or cross-sell agent?

Skip it when the cart already has a deterministic accessory rule that is correct (camera body → approved lens SKUs), when compatibility is not in the catalog, or when leadership wants silent cart-add for margin. Also skip the older Agents Classic product for new work after July 30, 2026. Missing “fits” data is a product information management (PIM) ticket, not a model job.

What could go wrong if we rank add-ons only by frequently bought together?

Incompatible accessories, duplicates the shopper already owns, and out-of-stock “bestsellers” land in the cart drawer. Co-occurrence is a tie-break after relationships, stock, and history pass. Same rule as the recommendation-agent post — cart-aware, not a different religion.

What could go wrong if the upsell agent can write prices?

It will cut the add-on to close the attach. That is a discount engine hiding in merchandising. No setPrice, no issueDiscount. Offers are a separate path with a person approving (human-in-the-loop). Instructions in the prompt do not authorize a write.

How is this different from a product recommendation agent?

Recommendations start from session intent (“wide trail shoe”). Upsell/cross-sell starts from the current cart plus relationships. Share catalog and inventory tools. Do not share one unbounded prompt that also mints coupons.

Should the agent auto-attach the add-on to the cart?

Not in week one. Return a structured next-best payload. The storefront confirms. A later cart-mutation tool needs a hard block, shopper identity, and a confirm. Quiet attach of a high-margin unrelated SKU is how you earn chargebacks.

Is there a native Shopify AgentCore connector for upsell widgets?

No. Relationship graphs live in the PIM or middleware you expose as signed-in lookups. Shopify apps that “do upsell” are not AgentCore. Treat the platform as a system of record behind your own API.