About the retailer
- Region
- United Arab Emirates
- Brands on the line
- 5+
- AI employees
- 4
- Languages
- EN · AR · HI · UR · ML · DE · FR · ES
- Channels
- Voice, WhatsApp follow-up
- First go-live
- 30 days
A global retail and home appliance group in the UAE now runs an AI call centre for after-sales service: the work of its after-sales call centre moved onto four xAIa AI employees. The inbound AI employee knows the caller before it answers, qualifies the fault, registers the service case in the group's CRM and assigns a technician, handling more than 20,000 calls a month, 72% of them with no human escalation needed. It has already registered more than 1,000 service requests. Three outbound AI employees call dealers to collect complaints, rebook customers who missed a technician visit, and offer extended warranties to customers whose cover is ending. Every action is written into the CRM through xAIa's Agent API.
About the retailer
The client is a global retail and home appliance group with a large footprint in the UAE. It sells appliances and electronics through its own stores and a wide network of dealers, and runs its own after-sales operation: service centres, a call centre and field technicians who visit homes to install and repair air conditioners, washing machines, refrigerators, dryers and dishwashers across the country.
Its after-sales call centre is where all of that meets the customer. Its agents took every repair request, status question and complaint, and also had to call out to dealers, to customers who missed a visit and to customers whose warranty was ending.
Why was the after-sales call centre under pressure?
Summer peaks
When temperatures rise, air conditioners fail, and the lines that should be booking technicians fill up with calls about bookings already made.
Manual entry on every call
Agents switched between the phone and the CRM to find the customer, check the product and warranty, and file the case into fields with thousands of possible values. Calls were slow, and the service team had errors to fix later.
Outbound always came second
Dealer follow-ups, rebooking missed visits and warranty renewals depended on whatever time inbound left over, which in peak season was none.
Revenue leaking quietly
Missed visits that were never rebooked, dealer complaints that aged before reaching the service team, and warranty expiries that passed without a single conversation.
The AI workforce: four AI employees, one after-sales operation
Each AI employee has one job, its own permissions in the CRM and its own measures of success. They were deployed one at a time, each tuned on real calls before the next one started.
Inbound · Live
The inbound service AI employee
Answers every call to the after-sales line in the caller's language, and files the case before the call ends.
- Knows the caller before hello, from their CRM record: products owned, warranty status and past requests.
- Qualifies the fault and its urgency the way a trained service agent does. A fridge that stopped cooling and an AC leaking onto the floor are different priorities.
- Registers the case with every field set correctly from more than 2,500 options, and assigns a technician.
- Answers status questions from the live record, and hands complaints and anything outside its scope to the right team with the context captured.
- 20,000+ calls a month, 72% with no human escalation, and 1,000+ service requests registered so far.
Outbound · Live
The dealer care AI employee
Covers the retail channel: power retailers, chains and independent dealers who sell the group's brands.
- Calls each dealer on a set schedule.
- Collects new customer complaints raised in store and registers each one as a CRM case with the customer and product details.
- Updates the dealer on every open complaint from the live record, so nobody chases the service team by phone.
- Escalates urgent cases to the service team the same day.
Outbound · Live
The appointment rescheduling AI employee
Turns missed technician visits back into completed repairs.
- Calls every customer whose visit was missed, using the job record in the CRM.
- Offers the next open technician slots and books the new visit straight into the system.
- Updates the job with the reason for the missed visit and the new slot, so the field team arrives to a confirmed appointment.
Outbound · Live
The extended warranty AI employee
Turns warranty expiries into new revenue.
- Calls customers whose standard warranty has ended or is about to, from a list built and screened by the client.
- Names the customer's own appliance and brand, quotes the exact plan prices supplied for that customer, and explains the difference between plans.
- Tells the customer who the payment goes to, based on the brand, and sends the brand's warranty catalogue on WhatsApp during the call.
- Writes the decision into the CRM. An accepted plan becomes a call centre task with the plan, price and confirmed messaging number filled in.
Agent API: how AI employees work inside the retailer's own systems
A voice agent that only talks is a nicer IVR. The value is in what happens in the system while the customer is still on the line. Agent API is xAIa's integration layer that lets AI employees read and act inside a company's own systems, with the same accuracy and the same permissions as a trained member of staff.
Reads before the call
Matches the caller's number to the customer, their products, warranties and open cases before the AI employee says hello.
Writes during the call
Registers service cases, assigns technicians, creates call centre tasks, rebooks visits and updates lead statuses in real time.
Turns talk into clean data
Maps every answer to the right value among more than 2,500 field options, so the service team receives cases it can act on without retyping.
Enforces the rules
Sets a Do Not Call flag the moment a customer asks, and a flagged customer is never called again until the client's team removes it.
Logs everything
Attaches each call to the CRM record with the recording, the transcript, the outcome and every action taken.
Production grade
Separate staging and production, scoped permissions, more than 1,600 automated tests on every release, and one switch to stop all writes instantly.
The service team works in the same CRM it always has. The difference is that the cases now arrive complete, correct and assigned.
How a service call runs
1
Before hello
The number is matched to the customer record, products and history.
2
On the call
The AI employee confirms the product, qualifies the fault and its urgency, and agrees the next step with the customer.
3
Before goodbye
Agent API registers the case with every field correct and assigns a technician. The customer hangs up with a case reference.
Built for an enterprise call centre
- The AI employees answer on the group's own numbers and extensions, so customers and dealers call the same lines as before.
- They speak English, Arabic, Hindi, Urdu, Malayalam, German, French and Spanish, switching when the caller does.
- Every outbound call opens with a recorded-line disclosure, calling hours are kept, and every list is screened against Do Not Call before it is dialled.
- The client signed off every step: a solution architecture before build, user acceptance testing on real calls, and a 30-day hypercare period per agent.
- Prices, plans and policies come only from the client's own data.
What changed for the retailer?
- 20,000+ inbound service calls a month handled by AI
- 72% of calls need no human escalation
- 1,000+ service requests registered in the CRM so far
- 2,500+ CRM field options matched automatically
- 4 AI employees across the after-sales journey, 1 inbound and 3 outbound
- 30 days to the first agent live
The call centre team now supervises AI employees and handles the conversations that need judgement. Every service request, dealer complaint, rebooked visit and warranty decision lands in the CRM as structured data, so volumes, outcomes and trends can be read straight from it.
Frequently asked questions
Can one AI employee take service calls for several appliance brands?
Yes. At this retailer one inbound AI employee takes service requests for every brand on the line, with each brand's products, service options and rules loaded.
Can an AI voice agent register cases directly in our CRM?
Yes. Through xAIa's Agent API the AI employee registers the case while the customer is on the call, sets every field from more than 2,500 options and assigns a technician.
How does the AI employee know who is calling?
The caller's number is matched to the customer record before the call is answered, so the AI employee already knows the customer's products, warranty and open requests.
Can AI make outbound calls for extended warranties in the UAE?
Yes. The extended warranty AI employee calls customers whose cover is ending, quotes their exact plan prices, sends the catalogue on WhatsApp and records the decision in the CRM. Lists are screened against the Do Not Call Registry, each call opens with a recorded-line notice, and any customer who asks is never called again.
Can AI rebook missed technician appointments?
Yes. The rescheduling AI employee calls customers whose visit was missed, offers the next open slots and books the new visit in the system.
How long does it take to go live?
The first AI employee went live in 30 days, including CRM and telephony integration. Each further agent followed with its own testing and 30-day hypercare period.
