AI COMMERCE AGENT

Turn every conversation into one buying journey.

AI agents help customers find the right product, check stock, complete the order and get support afterwards — consistently across web, Zalo, WhatsApp and social.

Book a workshopSee a real scenario

Before the purchase

Product discovery

Trong mua

Cart & Checkout

Sau mua

Fulfillment & Care

GIAO DIỆN MINH HOẠ

The gaps in the buying journey

Revenue is not only lost at checkout.
It leaks away at every touchpoint.

Customers can leave before buying, during checkout or after delivery — simply because nobody helped at the right moment.

A retail assistant advising a customer and packing an order at the counter

An illustrative commerce journey

01 / 04

The moment a customer needs guiding

Before they leave the site or switch to another brand.

01

Bottleneck · Product discovery

Customers see plenty of products but cannot tell which one to pick.

A large catalogue and static filters cannot understand needs like “light”, “easy to match” or “good for beginners”. Customers are left to hunt, and often leave before a basket exists.

Common signalsSearches with no results · filtering over and over · asking the same question again
The customer is searching for“light running shoes for beginners”0 results
Search sessionThe customer is about to leaveWhere it leaks
The Commerce Agent can step in hereAsk what they need, understand the budget and priorities, then check the catalogue and stock to offer options that fit.
One agent · the whole journey

Alongside the customer before the purchase.
Support continues after the order arrives.

The Commerce Agent holds the same data, policies and customer context across all three stages — nothing restarts at each touchpoint.

01 · Product discovery

Turn a need expressed in plain language into the right product choice.

The agent asks for what is missing, checks the catalogue and stock, then explains briefly why each product fits.

What the agent takes on

Understand the need and the budgetWithout making the customer work through a wall of filters.
Check SKUs and stockOnly suggests products whose data has been approved.
Explain and compareSo the customer feels sure before adding to the basket.
Operational resultsA clear need, a matching list of SKUs, and enough context to move on to the basket.
AI Product FinderBefore the purchase · Running

Hand over to staff

Customer · Website“I need light running shoes for a beginner, under about two million.”
What the agent understood5–10 kmBeginner< 2 triệuLightweight priority
Suggested from the catalogueThe 2 closest matchesChecked against stock

AirFlow Runner 2

Light · Cushioned · Easy to get used to

1.790.000đ

96% match

Flex Pace Pro

Stable · Breathable

1.950.000đ

91% match

The AirFlow Runner 2 is lighter and suits beginners. Shall I check your size and the nearest branch?

Compare in detail

Check size 39

The context is kept

Need5–10 km
BudgetUnder two million
PriorityLight · easy to get used to
CatalogStockCRMGIAO DIỆN MINH HOẠ

Context is not lost when the stage changes

Need and choiceSaved from product discovery
Basket and paymentContinues with the same SKU and policy
Orders and after-salesReuses the history to support after the sale
Connecting the commerce stack

The agent does not sit outside your systems.
The agent works with the data you already have.

From catalogue and stock to orders, shipping and CRM — each connection states exactly what data is read, which actions are allowed, and which steps need approval.

System layers

01 · Catalog & Commerce

Read the right product. Apply the right policy.

Connected within permissions

Shopify

Catalog · Product API

Public API available

WooCommerce

REST API

Public API available

Haravan / Sapo

API depends on the system

Needs technical assessment

In-house commerce

Custom API / Webhook

Needs technical assessment

Commerce Agent

Product & Order Agent

Read scope: Catalog
Write scope: Draft only

Guardrails active

Search and filter SKUsAttributes · Variants · Collections
Check availabilityPrice · Stock · Branch
Create a draft basketOnly once the customer confirms
Apply policyApproved vouchers and prices only

Permissions and actions

Policy: Commerce Agent v2

ReadableName, description, variants, public price, available stock
Approval requiredChanging prices, discounts outside policy, refunds
No accessCost prices, full payment details, anything outside scope
10:42:18Read 24 SKUs from the “Running” collectionShopify Product API · via the Product & Order AgentRead onlyAudit log
Every business runs a different commerce stack.Actual connectivity depends on the APIs, permissions, data architecture and security requirements of the systems you already run. The platform names above only illustrate the direction of an API integration.
Assess your systems
Illustrative scenario

One conversation.
One uninterrupted buying journey.

The customer does not know what to choose. The Commerce Agent understands the need, checks real data and supports them until the order arrives.

R

Urban RunIllustrative retail brand

Hypothetical scenario

“I need running shoes for 5–10 km, I have wide feet, and my budget is under two million.”

AN

New customerCame from a Facebook ad
BeforeStaff ask the same questions repeatedly, check stock by hand, and lose customers when replies are slow.

The Commerce Agent handling it

From need to order — in one flow

Live scenario

Understand the need5–10 km · wide feet · under two million12 seconds
Find the right productFilter the catalogue, sizes and stock by branch3 options
Create the basket and confirmValid offers applied, customer confirms size 421 draft order
Follow up after the purchaseShipping updates and guidance on changing sizeAutomated
First response< 15 giâyinstead of waiting for staff
Manual work4 → 1only exceptions need approval
StatusDeliveredwith the history of the whole journey
The figures in this scenario illustrate how it operates; they are not a commitment on results. Actual performance depends on your data, processes and connected systems.
See the agent handle your own scenario
Frequently asked questions

Before you let an agent into the sales journey.

What businesses usually want cleared up about data, permissions, integrations and rollout.

Your question is not here?The InCard team will work through it around your actual systems and processes.
Talk to an expert

No. The agent takes the repetitive work — common questions, finding products, checking status and drafting orders. Staff still handle in-depth advice, exceptions and any decision that needs approval.

The agent supports operations — people keep the decisions.

Next step

Start with one specific commerce scenario.
You do not have to replace the whole system.

A workshop to review your buying journey, pick the priority scenario and work out what data to prepare before integration is discussed.

Three inputs to get started

Commerce platformThe website, marketplace or sales system you run
Order and stock systemsWhere your stock data and order status already live
Priority use caseThe revenue leak you want to tackle first
The workshop does not require connecting to live systems. Integration scope is agreed only after reviewing the APIs and permissions.