Customer support · Production

Swiggy builds generative AI neural search, conversational bots, and LLM-powered restaurant partner self-service

The problem

Users find it daunting to choose from the many food options on Swiggy's app, unfamiliar dish names create confusion, and restaurant partners lack efficient self-service for onboarding and operational questions.

Workflow diagram · grounded in source
1
User submits open-ended query
trigger
“Swiggy's neural search enables users to search using conversational and open-ended queries and receive recommendations tailored to their specific needs. This makes it easier for consumers to find what they're looking for without having t…”
2
LLM processes food search
ai_action
“built using a Large Language Model (LLM), adapted to understand the terminology related to dishes, recipes, and restaurants and Swiggy-specific search data. We have 50 million-plus items in our food catalog covering a wide variety of opt…”
3
Personalized recommendations delivered
output
“receive recommendations tailored to their specific needs”
4
Catalog enriched with AI content
ai_action
“we have leveraged generative AI techniques to enrich our catalog with images and detailed descriptions of items”
5
GPT-4 chatbot handles customer queries
ai_action
“we are collaborating with a third-party to develop a GPT-4 powered chatbot. This bot will enable efficient and empathetic service to oft-asked customer queries”
6
LLM enables restaurant partner self-service
ai_action
“in-house tuned LLMs to empower restaurant partners to self-serve on processes and questions related to onboarding, ratings, payouts, etc, leading to faster issue resolution and streamlining. A conversational assistant powered by this LLM…”
Reported outcome

Neural search enables conversational food and grocery discovery, a GPT-4 chatbot handles customer service empathetically, and in-house LLMs power restaurant partner self-service leading to faster issue resolution.

Reported metrics
Food catalog size50 million-plus items
Partner issue resolution speedfaster issue resolution and streamlining
Reported stack
LLMGPT-4WhatsApp
Source
https://bytes.swiggy.com/swiggys-generative-ai-journey-a-peek-into-the-future-2193c7166d9a
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Neural search enables conversational food and grocery discovery, a GPT-4 chatbot handles customer service empathetically, and in-house LLMs power restaurant partner self-service leading to faster issue resolution.

What tools did this team use?

LLM, GPT-4, WhatsApp.

What results were reported?

Food catalog size: 50 million-plus items; Partner issue resolution speed: faster issue resolution and streamlining (source-reported, not independently verified).

How is this customer support AI workflow structured?

User submits open-ended query → LLM processes food search → Personalized recommendations delivered → Catalog enriched with AI content → GPT-4 chatbot handles customer queries → LLM enables restaurant partner self-service.

Related customer support cases