Use cases/Retail & E-commerce
Order status, returns and refunds
Answers "where is my order", processes returns and screens damage photos on every channel.
WhatsAppWebsiteVoiceEmailReads documentsLive in about 3 weeks
- 24/7 replies
- Peak season without hiring
- Refund abuse stopped at intake
How it runs
78%of contacts handled start to finish by AI employees
- Every month
- 1,50,000
- contacts
- Your team steps in
- Approval checkpoint
- 4 hours
Illustrative. A 30-day pilot measures your own baseline.
One call, start to finish
A damaged mixer grinder, returned and refunded
On WhatsApp
One call, start to finish
A damaged mixer grinder, returned and refunded
On WhatsApp
Damaged orderA refund asked for
Customer
My mixer grinder arrived broken. I want a refund.
A damaged item and a refund, understood in the first message.
Case record
- 01Complaint loggedOrder 8842
- 02Return eligibleWithin 7 days
- 03Photos screenedGenuine
- 04Refund approved₹3,499
- 05Pickup bookedTomorrow
One call, start to finish
A damaged mixer grinder, returned and refunded
On WhatsApp
01Damaged orderA refund asked for
Customer
My mixer grinder arrived broken. I want a refund.
WhatsApp
A damaged item and a refund, understood in the first message.
02EligibilityOrder and returns policy
AI employee
Sorry about that. Order 8842 can be returned. Could you send a photo of the damage?
Order systemReturns policyYour SOP
Eligibility is checked against the order and your returns policy.
03Photo screeningChecked for reuse or edits
Customer · Here are the photos.
AI employee
Thank you. The photos look genuine. I've passed the refund for approval.
Photo screeningDocument intelligence
Every photo is screened for reuse, cloning or AI generation.
04Refund approvalAbove ₹2,000, a CX lead
CX lead
Refund of ₹3,499 approved.
Your team
Refunds above ₹2,000 need a CX lead's yes.
05Pickup bookedRefund confirmed
On record
Pickup booked for tomorrow. Refund confirmed on WhatsApp.
Chat keptPhotos storedApproval loggedAudit trail
Every refund, with its photos and its approval.
Case record
- 01Complaint loggedOrder 8842
- 02Return eligibleWithin 7 days
- 03Photos screenedGenuine
- 04Refund approved₹3,499
- 05Pickup bookedTomorrow
The workflow
Your SOP, run step by step.
6 steps across your customer, AI employees, your team and your systems. 78% of cases run start to finish without a person; your team steps in at one point, only when the SOP says.
AI employees do the workYour team, only when the SOP saysReads and checks documents
Your SOP and your people
Written once by you. Followed on every case.
The SOP, as writtenPlain words, the way your team already works
Answer order-status questions from the order system. For returns, check eligibility and book the pickup. If the customer sends damage photos, screen them. Refunds above ₹2,000 or with a high fraud score need a CX lead's approval.
Document intelligence: Screens damage photos for reuse, cloning or AI generation.
Where your people step inApproval checkpoint
- Who
- CX leads group
- When
- Refund above ₹2,000, high photo-risk score, repeat refund requester
- Within
- 4 hours
- They see
- Order, photos with heatmap, customer refund history
Measured
Every case measured. Better every week.
Every conversation is reviewed overnight. What it finds comes back as a change for your SOP owner to approve.
Of every 100 contacts
Contacts100
Order identified95
Answered by AI80
Resolved first time78
No repeat contact in 7 days70
Illustrative share of a month's volume reaching each stage.
This week's suggestion · from the nightly review
'Delivered but not received' sent 1,900 chats to people. Proposal: ask for the courier's photo proof and open a claim automatically.
Nothing changes until your SOP owner approves it.
On the record, every case
- Every answer with the data it used
- Photo scores and reason codes
- Refund approvals with names
- Full conversation across channels
Voxket AI Office
Your AI office is ready when you are.
Bring one SOP. We'll show it running.
Go beyond intelligence.