Swiggy expands gen AI application from just food to q-commerce, dining out

The company feels the move is required as the platform has over 50 million items as part of its food catalogue and looks to change the way how consumers place orders.

Swiggy on July 20 said it will now be using generative AI (gen AI) to ready products and services that will boost food delivery and dining out experience.

Earlier in the week Chief Executive Officer of the company’s food marketplace Rohit Kapoor told Moneycontrol that the startup has formed a dedicated team of around five members, who are tasked with finding new applications of artificial intelligence (AI) throughout the company as it looks to better efficiency and lower costs.

The company feels the move is required as the platform has over 50 million items as part of its food catalogue and looks to change the way how consumers place orders.

“Imagine being able to type queries like, ‘I just finished my workout. Show me healthy lunch options,' or 'Show me vegan-friendly starters’. With Swiggy's neural search, these open-ended, conversational queries are (now) understood, and personalised recommendations (will be) provided,” Madhusudhan Rao, CTO at Swiggy, said in a blog post.

He added that neural search capabilities have been built using a Large Language Model (LLM), which has been adapted to understand dish, recipe and restaurant terminology as well as Swiggy-specific search data.

That way consumers will be able to find what they're looking for without having to use or remember specific keywords.

Those capabilities were possible even for Swiggy Instamart, its quick-commerce unit and dineout as well.

“Our unique Dineout conversational bot acts as your virtual concierge, guiding you to restaurants that meet your preferences be it ambience, kid-friendliness, valet parking, ratings, cost, to name a few. As part of our commitment to improving customer service, we are collaborating with a third party to develop a GPT-4 powered chatbot,” Rao concluded.

Swiggy’s neural search feature will be beta-tested by September and based on the learning and results, the company hopes to roll out the feature to all search traffic in the app.