Skip to content

AI agents for eCommerce

Resolve routine customer questions and escalate the exceptions.

Where-is-my-order, sizing, returns and product questions make up most of an online store’s inbox. An AI agent answers them from your policies and live order data on web chat, WhatsApp and email, processes what it is permitted to, and hands the unusual cases to your team with the context attached.

Book a Free DemoSee the Customer Support Agent

Free 30-minute consultation · No obligation · Built around your workflow

Two people packing online orders at a long workbench in a bright fulfilment studio

Sound familiar?

Where enquiries slip through today.

The everyday friction teams describe before an AI agent takes the first response.

  • “Where is my order?” is most of the inbox, and every reply means looking up the same information.

  • Returns and exchanges follow clear rules, yet each one is handled by hand.

  • Product and sizing questions go unanswered at night and on weekends—when many people shop.

  • Complaints, damaged items and payment issues get buried under routine tickets.

  • Support, order and inventory systems do not talk to each other, so agents copy and paste between tabs.

Three high-value workflows

What the agent takes on, step by step.

Each workflow is configured around your rules, wording and systems. The steps show the typical shape; the details are yours.

  1. Order status and delivery questions

    Answer the most common question instantly, from live data.

    1. Identify the customer and order using the verification steps you define
    2. Look up order status and tracking in your store or fulfilment system
    3. Explain delivery windows, delays and next steps from your approved wording
    4. Update the address or delivery instructions where your rules allow
    5. Create a ticket and notify the team when a parcel is lost or damaged

    Outcome: Routine order questions resolved without a human touch

  2. Returns, exchanges and refunds within policy

    Apply your returns policy consistently and start the process in one conversation.

    1. Check the item and order date against your returns policy
    2. Explain the options—return, exchange or store credit—that apply
    3. Generate the return label or instructions and log the request
    4. Escalate refunds above your threshold, or outside policy, for human approval
    5. Update the order and notify the customer at each step

    Outcome: Returns handled by the rules, with exceptions approved by people

  3. Product questions and pre-sale support

    Help shoppers decide, with accurate product information.

    1. Answer sizing, material, compatibility and stock questions from your product data and approved content
    2. Suggest alternatives when an item is out of stock
    3. Share shipping costs, delivery times and payment options
    4. Hand complex or high-value enquiries to your team with the conversation attached
    5. Log common unanswered questions so your product content improves

    Outcome: Fewer abandoned carts caused by unanswered questions

Example conversation

Delivery query resolved, exception escalated

An illustrative conversation showing the agent at work—and what it does in your systems while it talks. Names, times and figures are examples, not client data.
  1. Customer: Hi, my order hasn’t arrived and it said 2–3 days. Order number 48213.
  2. Agent: I’m sorry it’s late—let me check. I’m the store’s AI assistant. To confirm it’s your order, could you tell me the email address or postcode on the order?
  3. Customer: Postcode 3036.
  4. Agent: Thank you. Order 48213 was dispatched on Wednesday and the courier’s last scan was Friday at their local depot. It hasn’t moved since, which is longer than usual. I can open a courier investigation and ask our team to contact you tomorrow morning. If the parcel is confirmed lost, would you prefer a replacement or a refund?
  5. Customer: A replacement, please. I need it for next weekend.
  6. Agent: Noted—replacement preferred, needed by next weekend. I’ve opened a ticket for the team with the courier details and your preference; they’ll confirm the next step by email tomorrow morning. Is there anything else I can check for you?
  7. Customer: No, that’s all. Thanks.
  8. System: Order verified · Courier investigation opened · Escalated to the team with replacement preference

What the agent does

Web Chat

Behind the conversation, in your systems.

  1. Verified the customer

    Order number matched with postcode

  2. Checked order and tracking

    Dispatched Wednesday · Last scan Friday, depot

  3. Applied the delivery-exception rule

    No movement for 48 hours → investigation

  4. Opened courier investigation

    Ticket #5521 with tracking history

  5. Drafted replacement request for approval

    Replacement preferred · Needed by next weekend

  6. Escalated to the support team

    Priority: high · Follow-up by 10:00 Monday

  7. Updated the customer record

    Conversation and ticket linked to the order

Before and after

The same enquiries. A different day.

What changes when the agent handles the first response and the admin around it—and where your team stays in the loop.
Before

Manual first response

  • Most tickets are “where is my order?”, each answered by looking up the same screens.

  • Returns are processed by hand, one email thread at a time.

  • Pre-sale questions go unanswered outside office hours.

  • Damaged, lost and payment issues sit in the queue behind routine tickets.

  • Agents copy information between the store, courier and helpdesk tabs.

After

With an AI agent

  • Order status, delivery and policy questions are answered instantly from live data.

  • Returns and exchanges within policy are started in one conversation; exceptions go to a person.

  • Shoppers get product and sizing answers at any hour, in the channel they used.

  • Exceptions are recognised, prioritised and escalated with full context.

  • Your team handles judgement calls and relationships, not repetitive lookups.

Recommended package

The package we usually recommend.

Online stores usually need web chat, WhatsApp and email working together with live order data, returns rules and helpdesk integration—often through your store platform’s API. Business includes multiple channels, custom integrations, approval rules and reporting. Smaller stores with one main channel and no order lookups can start with Starter or Pro.

Agents involved

Also relevant

Business

Multiple agents and channels, including voice, with custom integrations.

One-time setup
From €3,990
Monthly plan
From €499/month
  • Multiple channels and workflows
  • Voice, WhatsApp, email and web
  • Custom CRM/API integrations
  • Advanced routing and approvals
  • Reporting dashboard
  • Priority support
Design My SolutionCompare all packages

Prices are starting points and exclude third-party usage fees such as AI model, WhatsApp, telephony, email, and external software charges. Final pricing depends on workflow complexity, integrations, volume, languages, and support requirements.

Human control

What the agent does—and what stays with your team.

The agent answers from your policies and approved content, looks up orders after the verification steps you define, and performs only the actions you permit—such as generating a return label or updating delivery instructions. Refunds above your threshold, goodwill gestures, complaints, chargebacks and anything outside policy are escalated for human approval. It does not handle card details; payments stay in your payment provider’s systems. Consumer-rights wording follows what your legal advisers approve for the markets you sell in.

The agent

  • Collects information
  • Answers approved general questions
  • Schedules, routes and escalates

Your team

  • Makes every medical, legal, financial or eligibility decision
  • Approves anything outside the rules you set
  • Reviews the conversation history at any time

FAQ

Common questions

Honest answers about scope, data and control for this sector.

Shopify, WooCommerce and similar platforms offer APIs we connect to for order status, tracking and returns. Custom or headless stores are integrated through their own APIs. We confirm the exact scope after seeing your setup.

Only within limits you set—for example, automatic approval for returns within policy up to a value you define, with anything above that or outside policy sent to your team for approval. Many stores choose to keep all refunds human-approved at first.

It asks for details that match the order—such as the order number plus the email or postcode—before sharing anything. You choose the verification level, and it never asks for card details.

Yes. Email enquiries can be answered automatically or drafted for a human to approve, depending on the topic. Chat and WhatsApp are handled in real time, and all channels share the same knowledge and rules.

The agent recognises the complaint and frustration signals you define, stops trying to resolve automatically, acknowledges the issue in approved wording and escalates to your team with priority and full context.

It can suggest alternatives and answer questions using your product data and approved content. It does not invent product claims, and it flags questions your content does not answer so you can improve it.

Show us one enquiry. We’ll show you the agent.

Bring a real enquiry from your business to a free 30-minute session. We’ll map how an agent would handle it—and where your team stays in control.

Book a Free DemoMessage Us

Free 30-minute automation assessment.