Order Status and Returns Voice Agent
A multilingual voice agent that answers where-is-my-order calls from live courier data, starts returns and exchanges under your policy, and hands disputes to an agent with context.
- Capability:
- Voice AI
- Industry:
- Retail, D2C & Quick Commerce
Who it is for
- Head of Customer Experience
- Support Operations Manager
- D2C Founder or COO
- Returns and Reverse Logistics Lead
The problem
For most online sellers, where-is-my-order is the largest single reason customers get in touch, followed by returns, exchanges and refund status. Agents answer by switching between the OMS, courier panels and the payment gateway, while customers in smaller towns prefer to call and speak in their own language.
WISMO fills the queue
Simple status calls crowd out the complaints and disputes that genuinely need a person.
Too many screens
Agents look up the OMS, courier tracking, reverse pickup status and refund status separately to give one answer.
Return policy applied unevenly
Rules on non-returnable categories, return windows and exchange stock are interpreted differently by different agents.
Language and channel preference
Many customers would rather call in Hindi, Tamil, Bengali or another language than use chat or email.
What we build
We build an inbound voice agent connected to your OMS, courier tracking and payment gateway. It verifies the caller by registered number or OTP, finds the order and answers status questions from live tracking, including delays and failed delivery attempts. For returns and exchanges it calls your policy engine, which checks category, window, condition and exchange stock. It books the reverse pickup, explains the refund timeline and sends a WhatsApp confirmation, in Hindi, English or a regional language. Disputes, damaged-item complaints, high-value orders and frustrated callers go to a human agent with a call summary.
- Verifies callers by registered number or OTP before sharing order details
- Answers status from live OMS and courier data, including delays, failed attempts and rescheduling
- Starts returns and exchanges through your policy engine, including category, window and stock rules
- Books reverse pickups and gives refund status from the payment gateway or wallet
- Speaks Hindi, English, Hinglish and regional languages, with WhatsApp follow-up messages
- Hands over to a human agent with a call summary for disputes, damage claims and escalations
How it works
- 01Verify the callerThe agent matches the calling number to orders or sends an OTP, then confirms which order the call is about.
- 02Understand the requestThe caller explains in their own words, and the agent classifies it as status, return, exchange, refund or other.
- 03Fetch live dataOrder, shipment, pickup and refund status are fetched from the OMS, courier and payment systems.
- 04Apply policyFor returns and exchanges, the policy engine decides eligibility and the agent explains the result.
- 05Act or hand overEligible requests are booked; disputes, damage claims and escalations go to a human agent with a summary.
- 06Confirm on WhatsAppThe customer receives a confirmation with the pickup window or the next update.
Architecture
- Telephony
- Support number on cloud telephony or a contact centre platform
- Call recording with disclosure
- Conversation
- Multilingual speech-to-text and text-to-speech
- LLM dialogue manager with intent and slot schemas
- Escalation and frustration rules
- Commerce data
- OMS order and return APIs
- Courier and aggregator tracking APIs
- Payment gateway refund status
- Policy
- Return and exchange policy engine
- Customer verification service
- Handover and audit
- Agent desktop handover with summary
- Transcripts with PII redaction and QA sampling
Inputs and outputs
Inputs
- Inbound calls to the support number
- Order and return data from the OMS
- Shipment tracking from couriers or aggregators
- Refund status from the payment gateway
- Return and exchange policy by category
Outputs
- Answered status calls and booked returns or exchanges
- Reverse pickup requests in the OMS or courier system
- WhatsApp confirmations to customers
- Human handovers with call summaries
- Daily report of intents, languages and handover reasons
Where people stay in control
Disputes, damaged or wrong item claims, high-value orders and any caller who asks for a person go to a human agent with a summary. Refunds outside policy are decided only by your CX team.
Integrations
- OMS such as Unicommerce, Shopify or an in-house system
- Courier and shipping aggregator APIs
- Payment gateway refund APIs
- Helpdesk such as Freshdesk or Zendesk
- WhatsApp Business API
Guardrails
- Caller verification before any order detail is shared
- Return eligibility decided by the policy engine, not the language model
- Delivery dates quoted only from courier data
- Immediate handover on request or detected frustration
- Card and bank details never collected by voice
Compliance notes
- Return, refund and grievance information follows your published policy, in line with the Consumer Protection (E-Commerce) Rules 2020.
- Call recordings and transcripts are personal data under the DPDP Act 2023; PII is redacted in logs and retention is limited.
A typical pilot
Duration: 4 weeks
Status and return-initiation calls for one brand or channel in Hindi and English, with overflow and after-hours calls routed to the agent first.
Deliverables
- 01Call flows and policy rules agreed with CX
- 02Working agent on a pilot number or overflow queue
- 03Integrations with OMS, courier tracking and refund status
- 04Pilot report on resolution, accuracy and handovers
- 05Roll-out plan with language additions
What we measure
- Resolved without handover
- Calls completed by the agent with no repeat call about the same order within the agreed window.
- Policy decision accuracy
- QA review of a sample of return and exchange outcomes against your policy.
- Handover reasons
- Logged reason for each transfer, reviewed weekly with CX.
- Customer satisfaction
- Post-call survey compared with human-handled calls of the same type.
FAQs
01Will customers know they are talking to an AI?
Yes. The agent introduces itself as an automated assistant, and callers can ask for a person at any point. Clear disclosure also reduces complaints.
02How does it handle angry customers?
It listens for frustration and requests for a person, then hands over with a summary so the customer does not repeat themselves. It does not argue or try to keep the caller.
03Can it process refunds?
It reports refund status and books returns that your policy engine approves. Refund exceptions are decided by your CX team.
04Which systems do you need access to?
Read access to orders, shipments and refunds, and write access to create returns and pickups. We use your existing APIs, or a middleware layer if your OMS has limited APIs.
Related use cases
- Retail, D2C & Quick Commerce
COD Order Confirmation Calls
Call every cash-on-delivery order soon after it is placed, in the customer's language, to confirm items, amount and address, offer a UPI prepaid option and tag the order before it ships.
- Retail, D2C & Quick Commerce
Dark-Store Picking and Packing QC
A camera at the packing station checks each tote against the order before the bag is sealed, catching wrong variants, missing items and near-expiry stock, and keeps an image per order.
- Logistics & Supply Chain
Delivery Status Agent (Voice + WhatsApp)
Answer where-is-my-shipment calls and WhatsApp messages from consignees and shippers, arrange reschedules and NDR reattempts within your rules, and hand complaints and disputes to your team.
- Fintech & Payments
Support Voice Agent for Failed Payments and Refunds
A Hindi, English and Hinglish voice agent that finds debited-but-failed UPI and card payments, explains reversal and refund status from live data, and raises or escalates tickets.
Founded by Akhilesh Mittal. 8 years building a security platform for a New Zealand company. IIT ISM Dhanbad and IMT Ghaziabad. Builds production agentic AI.
Pick one leak and measure it
Book a discovery call. We will choose one narrow pilot, such as COD confirmation, one claim type or one dark store, run it alongside your current process and report what we measured.