Flipkart
conversational genAI shopping assistant
Launched in 2023, Flippi, Flipkart's conversational genAI shopping assistant, runs in production with 1.3 million engaged monthly active users, a last-touch conversion of roughly 1%, and 68% positive reviews.
Key points
- Conversational genAI shopping assistant (rewriting, intent, RAG, entities).
- In-house modular Flippi architecture backed by OpenAI, iterated through A/B testing.
- 1.3 million engaged monthly active users, roughly 1% conversion, 68% positive reviews.
- Evidence level B, confirmed active status.
Objective
Bring online shopping closer to the experience of an in-store salesperson: the customer describes what they want in natural language, and the assistant helps them discover, compare, and choose, to remove product discovery friction and convert more.
The deployment
Flippi is Flipkart's conversational genAI shopping assistant, launched on October 4, 2023 ahead of the major Big Billion Days event. The customer says what they are looking for and receives options to compare, as if talking to a salesperson. Under the hood, a modular architecture chains query rewriting, intent detection, RAG, entity recognition, and context reduction to handle product discovery, personalization, offer identification, and comparison. According to the technical paper published by the Flipkart teams, the system runs in production with A/B testing iteration cycles, tested on search, browse, and category pages. It reaches 1.3 million engaged monthly active users, a last-touch conversion of roughly 1%, and 68% positive reviews (thumbs up).
Results Proof B
Quantified technical paper published by the Flipkart teams (official interested source, production measurements and A/B tests), corroborated by established Indian press on the launch and positioning. The figures come from the brand, not a third-party audit, which places it at B.
How it works
Documented architectureThe stack in detail
- llm OpenAI Underlying generative model; the assistant was presented as ChatGPT-powered at launch, without the exact model version being published
- plateforme Flippi (architecture modulaire in-house) Pipeline built by the Flipkart teams: query rewriting, intent detection, entity recognition, context reduction
- outil Couche RAG sur le catalogue Flipkart Retrieval of products, offers, and comparisons from the indexed catalog to compose the responses
- outil Cadre d'A/B testing in-house Production iteration cycles on search, browse, and category pages, with conversion and satisfaction measurement
How it runs, concretely
For ops teams-
1Expressing the need customer
The customer describes what they are looking for in natural language in search or the assistant.
-
2Understanding the query AI
Rewriting, intent detection, and entity recognition structure the request.
-
3Retrieval and response AI
RAG retrieves products, offers, and comparisons from the catalog and composes the response.
-
4Iteration data team
The teams measure conversion, satisfaction, and session coverage and adjust through A/B tests.
The ability to answer the session and the conversion that follows. If the catalog and offers are not well indexed for RAG, the assistant gives vague or wrong answers.
How your customers perceive this type of use
Sourced studiesLes consommateurs n'acceptent pas les chatbots par defaut : 64% prefereraient que les entreprises n'utilisent pas d'IA dans leur service client (Gartner, 2024) et pres d'un utilisateur sur cinq du service client par IA n'en retire aucun benefice (Qualtrics, 2025). L'acceptation se construit sur trois conditions mesurees par Salesforce : savoir qu'on parle a une IA, pouvoir escalader vers un humain, comprendre la logique de l'agent.
Acceptance conditions
- Etre informe qu'on parle a une IA et non a un humain (pres de 75% le demandent, Salesforce 2024)
- Un chemin d'escalade clair vers un agent humain (45% plus enclins a utiliser l'agent IA, Salesforce 2024)
- Une logique de l'agent clairement expliquee (44% plus enclins, Salesforce 2024)
Red lines
- Rendre l'humain injoignable : c'est la premiere inquietude des consommateurs sur l'IA dans le service client (Gartner 2024) et 50% craignent que l'IA les coupe du contact humain (Qualtrics 2025)
- Remplacer le service client par l'IA sans alternative : 53% envisageraient de partir chez un concurrent (Gartner 2024)
Sources: Salesforce 2024 · Gartner 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- Rich, well-structured product catalog
- Offer and promotion base indexed for RAG
- Session history to measure coverage and conversion
Org prerequisites
- Production A/B testing framework
- AI transparency rule toward the customer
- Team able to maintain a modular pipeline (intent, RAG, entities)
Possible stack
- Generative LLM
- RAG layer over the catalog and offers
- Intent detection and entity recognition modules
The plan, step by step
- Step 1Index the catalog, offers, and product attributes for RAG; define the pilot scope on a high-traffic category.Deliverable: Queryable catalog index and validated pilot scope
- Step 2Assemble the pipeline: query rewriting, intent detection, RAG, entity recognition, with a market LLM via API.Deliverable: Working assistant in a test environment on the pilot category
- Step 3Launch in an A/B test on search and the pilot category pages, with a clear notice that the customer is talking to an AI.Deliverable: First production A/B test with conversion, satisfaction, and session coverage measurement
- Step 4Iterate on the uncovered cases (vague answers, poorly recognized entities), then extend to other categories.Deliverable: Quantified pilot review and per-category extension plan
First step: Cleanly index the catalog and offers for RAG, then launch a pilot on a high-traffic category with conversion measurement.
The same pattern, proven in other industries
Booking.com
genAI trip planning assistant
Burberry
genAI copilot for consultants (search, image recognition, complete-the-look)
Homes.com
natural-language voice and text search, then conversational assistant
Sources
- S1 Flippi: End To End GenAI Assistant for E-Commerce Interested party archive pending
- S2 Flipkart Launches ChatGPT-Powered Shopping Assistant 'Flippi' Established press archive pending
- S3 Access. Convenience. Safety: Leveraging technology to empower every Indian's dreams Interested party archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.