Flipkart Certified · 10 Cr+ Customers Served

Shopping Karo
Apni Bhasha Mein

AI-powered multilingual voice assistant for Flipkart — understands English, Hindi, and Hinglish with human-grade clarity. Trained on 50+ e-commerce intents including Orders, Returns, Delivery, and Seller Support.

Order TrackingReturn RequestDelivery IssueRefund StatusProduct QuerySeller Help
10 Cr+Customers Served
4.8MQueries Resolved/Day
96.4%Intent Accuracy
Live Voice Demo
Hinglish
U

Mera order kahan hai abhi?

"Where is my order right now?"

Listening
Understanding
Responding

Tap the mic to start the demo

Scroll to explore

Built for Flipkart Shoppers, Ground Up

Not a generic chatbot retrofitted for e-commerce. BhaashaSetu is trained exclusively on Flipkart's shopping workflows and Indian linguistic patterns.

Human-Grade Voice Synthesis

Neural TTS with prosody control, regional accent adaptation, and emotional tone modulation — indistinguishable from a trained Flipkart customer care executive.

WaveNet TTSProsody ControlAccent Adaptation

Hindi · English · Hinglish

Seamlessly transitions between languages mid-conversation. Understands code-mixed speech like "Mera order kab aayega?" without breaking context.

Code-Mixed NLPMid-Call Switch12 Dialects

E-Commerce Trained NLP Engine

Fine-tuned on 2.4M+ shopping conversations. Recognizes 50+ intents including Order Tracking, Returns, Refunds, Delivery Issues, and Seller Queries.

50+ IntentsEntity ExtractionFlipkart Vocabulary

24/7 Zero-Downtime Service

99.97% uptime SLA backed by multi-region deployment. Auto-scales during peak hours (Big Billion Days, sale events) without pre-provisioning.

99.97% UptimeAuto-ScaleMulti-Region

From Voice to Resolution

A 6-stage AI pipeline purpose-built for Indian telephony infrastructure, processing each customer query in under 2 seconds end-to-end.

01

Voice Capture

ASR Engine

Noise-robust speech recognition with 8kHz telephony support and VAD (Voice Activity Detection) optimized for Indian mobile networks

WebRTCVAD8kHz
02

Language Detection

LangID Model

Per-utterance language identification supporting Hindi, English, Hinglish with 99.2% accuracy across Indian regional accents

FastTextCode-MixHindi BERT
03

Intent & NER

NLP Pipeline

Transformer-based intent classification + named entity extraction (Order ID, Product Name, Amount, Date, Address)

IndicBERT50+ IntentsNER
04

Flipkart OMS API

Order Management

Secure integration with Flipkart Order Management System, Logistics API, and Seller Portal via encrypted REST endpoints

OMS APILogistics APISeller Portal
05

Response Gen

Dialog Manager

Context-aware response generation with multi-turn memory and slot-filling for complex queries like return + refund flows

LLM-HybridSlot FillContext
06

Voice Synthesis

Neural TTS

WaveNet-based TTS with regional accent profiles and real-time prosody adjustment for friendly, helpful customer care tone

WaveNetProsodySSML
< 2s
End-to-End Latency
3.2%
ASR Word Error Rate
96.4%
Intent Accuracy
4.3/5
TTS Naturalness MOS
50,000+
Concurrent Calls

Numbers That Matter to Flipkart

Aggregated across live deployments spanning Flipkart customer care, seller support, and delivery operations. Audited by Deloitte India, Q4 2025.

0%
Reduced Wait Time

Average customer support wait time dropped from 8 min to 48 seconds after deployment

0%
Cost per Resolution

Bot resolution costs ₹1.80 vs ₹6.40 for human agent — proven across 12 deployments

0s
Avg Response Time

End-to-end voice pipeline: ASR → NLP → OMS → TTS in under 2 seconds on 4G

0.00%
Uptime SLA

24/7/365 availability with multi-region failover. Handles Big Billion Days traffic spikes.

4.8M+
Queries/Day
10 Cr+
Customers Served
₹280 Cr
Support Cost Saved
4.7/5
Avg CSAT Score

Every Shopping Need, One Voice

Pre-built intent modules for every major Flipkart customer workflow — deploy in days, not months. Each module is optimized for Indian e-commerce patterns.

Order Tracking & Status

Customers can check order status, expected delivery date, and shipment location — in their language, without navigating the app.

Sample utterance:

"Mera order kahan hai? Kal delivery thi"

Order StatusDelivery ETAShipment LocationDelay Alert

Returns & Exchanges

Return initiation, pickup scheduling, exchange requests, and return policy queries — fully automated with Flipkart OMS integration.

Sample utterance:

"Yeh product damage hai, return karna hai"

Return RequestPickup ScheduleExchangeReturn Policy

Refund & Payment Issues

Refund status tracking, payment failure resolution, wallet credit queries, and EMI issue handling via voice with OTP authentication.

Sample utterance:

"Mera refund 7 din se pending hai"

Refund StatusPayment FailureWallet CreditEMI Issue

Delivery Support

Address change requests, delivery rescheduling, missed delivery follow-ups, and delivery partner contact — all via voice.

Sample utterance:

"Address change karna hai delivery ke liye"

Address ChangeRescheduleMissed DeliveryAgent Contact

Product & Account Queries

Product availability, size/spec queries, wishlist management, account issues, and password reset — with voice-based identity verification.

Sample utterance:

"Yeh phone ka 8GB RAM wala model available hai?"

Product InfoAvailabilityAccount HelpPassword Reset

Seller Support

Seller onboarding guidance, listing issues, payment settlement queries, and policy clarifications — multilingual support for Flipkart sellers.

Sample utterance:

"Mera payment settlement kab hoga?"

Seller OnboardingListing IssueSettlementPolicy Query

Trusted by Flipkart Teams

Real results from Flipkart's customer experience, seller support, and logistics teams.

We deployed BhaashaSetu for our Hindi-speaking customer base in UP and Bihar. The Hinglish comprehension is genuinely impressive — it handles order tracking and return queries our previous IVR failed on 45% of the time. First-call resolution went from 52% to 89% in 3 months.
Key result: 89% first-call resolution
PS
Priya Sharma
Head of Customer Experience
Flipkart — North India Operations
The seller support module saved us enormous manual effort during Big Billion Days. 28,000 seller queries handled over the sale weekend without a single human agent involved. Settlement and listing issue resolution was seamless.
Key result: 28K queries, zero agents
AM
Arjun Mehta
VP — Seller Experience
Flipkart Marketplace
Our tier-2 customers in Telangana and Karnataka were struggling with English-only support. BhaashaSetu's Telugu-accented Hinglish support was a game changer for delivery rescheduling and missed delivery follow-ups. CSAT jumped from 3.2 to 4.8 within 60 days.
Key result: CSAT 3.2 → 4.8
KR
Kavitha Reddy
Director — Last Mile Delivery
Flipkart Logistics (Ekart)

Built by People Who Know E-Commerce

A founding team with deep roots in Indian e-commerce, Indic language AI, and enterprise marketplace infrastructure.

AM

Arjun Mehrotra

Co-Founder & CEO

Ex-Flipkart Product, IIT Bombay CSE

E-Commerce StrategyProductCX Design
KS

Dr. Kavitha Subramaniam

Co-Founder & Chief AI Officer

PhD NLP — IISc Bangalore, Ex-Microsoft Research

Indic NLPSpeech AIIndicBERT
RD

Rohit Desai

CTO & Head of Engineering

Ex-Meesho Platform, BITS Pilani

Distributed SystemsE-Commerce InfraScale
SA

Sneha Agarwal

VP — Partnerships & Growth

Ex-Amazon India, IIM Ahmedabad

Enterprise SalesMarketplace GTMSeller Relations

Backed by

Flipkart VenturesSequoia IndiaBlume VenturesKalaari CapitalNASSCOM DeepTech