Typewise for retail and e-commerce

AI customer service for retail and e-commerce

Where AI-assisted service can fit a retailer, what each workflow needs from your systems, and what stays with your team.

Book a demo callSee how it works

Retail service is several different jobs. A shopper choosing between three products needs advice; a customer whose parcel has no delivery date needs an answer and someone chasing the supplier; a store expert messaging a regular client needs the brand’s knowledge at hand; a wholesale partner needs an order processed without re-keying. Treating all of them as “the chatbot” is how retail AI projects stall.

Typewise agents read and write in connected CRM, ERP and ITSM systems, sensitive actions wait for approval, hand-offs carry the full context, and every action is logged. Whether any workflow below can run end to end in your organisation depends on what your systems expose and on the data behind them, so each is presented as a candidate design to evaluate, with what it needs, not as a promised capability.

Candidate workflows to evaluate

Retail workflows to evaluate, one boundary at a time

Product advice and sizing

Candidate design

Answer compatibility, fit and “which one for me” questions from your product data and guidance, including advice that sends the shopper to a dealer rather than to checkout.

What it needs from your systems

Product attributes and guidance that are complete and current for the ranges in scope, including the attributes your website does not show; safety-relevant compatibility must exist as explicit data.

What stays with people

Advice stays off for ranges where the data is incomplete or stale. Compatibility is checked against explicit data, not inferred.

Preventive information

Candidate design

Report recurring questions per product and journey stage, so content gaps are fixed on the product page or at checkout instead of answered a thousand times.

What it needs from your systems

Conversation data that can be grouped by product and stage, and a content owner who acts on the reports.

What stays with people

Deciding what to change on the site stays with your product and content owners.

Order status and changes

Candidate design

Look up orders for verified customers, explain delivery stages, and handle the changes your policy allows.

What it needs from your systems

An authenticated customer session your order system recognises; order and carrier data that are populated for the cases that generate contacts; line-item status where cancellations are to be partial.

What stays with people

Address changes after dispatch, payment-method-specific rules and anything the carrier controls stay with people or with the customer’s own self-service.

Returns, repairs and claims

Candidate design

Check eligibility, issue return instructions, and collect the evidence a claim needs.

What it needs from your systems

Eligibility rules in a system the AI can query; a connected label step if labels are to be generated rather than described; a validation rule for customer-supplied evidence.

What stays with people

Claims that depend on evidence are validated, not just submitted; goodwill decisions above your threshold go to a person.

Store experts and VIP messaging

Candidate design

Give store teams the same product knowledge and order context the online team has, in the channels they already use with clients.

What it needs from your systems

The messaging channel connected and approved for staff use; the same knowledge and order sources the online team relies on.

What stays with people

The relationship stays with the expert; AI assists the message, it does not replace the person.

Wholesale and dealer requests

Candidate design

Answer order status, document copy and availability questions from trade customers, within their own entitlements and price visibility.

What it needs from your systems

Trade-customer entitlements and documents reachable by the AI; dealer stock data where availability is to be stated.

What stays with people

Dealer stock the brand cannot see is not promised; where availability is unknown, the design says so and routes to the dealer.

Operating boundaries

What your team keeps

  • Identity is verified by your systems, not asserted in the conversation: account data and account changes require an authenticated customer.
  • Every action the AI may take is granted per action as read, recommend, draft or execute, and enforced where the action happens, not only in the instructions.
  • Sensitive actions wait for a named approver; hand-offs carry the full context to a person who owns the next step, including outside service hours.
  • Every step is logged, so your team can audit what was looked up, what was proposed and what was done.
  • Policy decides, not the API: an action your systems could perform is still off the table if your policy says so, and the matrix records why.
How it is enforced

Permissions, approvals and an audit trail.

Agents read and write only in the systems and scopes you connect. Sensitive actions wait for approval, hand-offs carry the full context, and every action is logged. Hosting, access control and certifications are described on the security page.

Security and compliance →See how it works →
Rollout considerations

What a rollout has to take into account.

  1. Step 1

    Start with readiness, not with the bot

    Check what your order, product and carrier data actually contain for the cases that generate contacts: missing delivery dates, unmapped tracking events, incomplete attributes. Readiness gaps decide the first workflow more than ambition does.

  2. Step 2

    Pick one workflow with low consequences

    Order status and return eligibility are reasonable first candidates; product advice tends to wait for the catalogue to be ready. Run supervised drafts before anything executes.

  3. Step 3

    Fit the handoff to your staffing

    A small team cannot approve requests mid-chat. Design holds, timeouts and after-hours behaviour around the coverage you really have, including weekends when the warehouse still ships.

  4. Step 4

    Plan around peak season

    The weeks with the most value can be the worst weeks to onboard. Go live in a quiet period, with volumes and channel mix measured for both seasons.

Customer stories

Results from real deployments.

Published customer stories from retail and consumer brands; each describes its own setup and results, not the candidate designs above.

E-commerce & Retail
Lehner: AI-first customer experience across three brands
E-commerce
Brack.ch cuts response time by 19% with AI-powered customer service tools
Health & Wellness
Beurer: a 90% resolution rate and customer service at a new level
See all customer stories →

Read more articles

Which customer contacts should you prevent, not automate?
Customer Experience
Which customer contacts should you prevent, not automate?
What should a customer-service AI be allowed to do?
Automation
What should a customer-service AI be allowed to do?
Is your customer data ready for an AI agent?
Implementation
Is your customer data ready for an AI agent?

Good questions. Straight answers.

In a candidate design it answers from your product data and guidance, so the quality depends on what that data contains. Evaluating it means connecting the product attributes the website does not show and keeping advice switched off for ranges where the data is incomplete or out of date.

Only where your policy allows it and your systems expose the action. Address changes after dispatch are often routed through the carrier and kept out of AI scope; cancellations may need to be line-item decisions that distinguish physical and digital items and suppliers. Each action is granted separately after evaluation.

The handoff design is built around real coverage: cases are held with a clear promise outside staffed hours, and nothing depends on a mid-chat approval from someone who is not there. Scope is narrowed to what the team can supervise.

Evaluate it on your own orders and products.

Book a demo and we’ll walk through these candidate workflows against your shop, order and ticketing systems and the policies your team keeps.

Book a demo callRequest pricing