Perplexity CEO Aravind Srinivas tells The Verge that many people are using chatbots like regular search engines. It makes sense to offer information on things they look for, like restaurants, directly from the source. So it’s integrating Yelp’s maps, reviews, and other details in responses when people ask for restaurant recommendations.

“Our underlying motivation is that we started off as this tool that appealed to people doing fact checks and research. But research can be on anything, like what if you’re a coffee addict, you’d want to know where the good coffee shops are,” says Srinivas.

As The Verge’s David Pierce recently pointed out, how Perplexity mixes text and links when answering questions was already a selling point compared to other chatbots like Copilot, Gemini, or ChatGPT. Srinivas says that consideration is part of why the company eventually chose Yelp. Srinivas says Yelp’s data allowed Perplexity to offer a variety of ways to provide answers, like adding maps, photos, or even quote reviews. 

The responses include links to businesses’ Yelp pages so users can view more reviews or other details. Perplexity did not disclose the financial details of the deal and says it will not use Yelp’s data to train any model, as the company uses existing models like GPT and Claude 2 for the chatbot.

The connection is part of Yelp’s Fusion program licensing data to third parties, including BMW. “By integrating our highly trusted and rich local data into Perplexity, we’re able to provide their users with reliable local search results,” says Chad Richard, SVP of business & corporate development at Yelp, in an email to The Verge.

Perplexity is the latest AI company to cut a deal for a source of continually refreshed, up-to-date information, though Srinivas points out it is not partnering with Yelp but rather licensing its data. Google will reportedly pay Reddit $60 million a year to access real-time data and use Reddit data to train its AI models. OpenAI signed agreements with news publishers like Axel Springer to get data for training. 

Other chatbots can also suggest cafes and restaurants, but much of this data comes from information models learned from the internet, almost like it did the searching for users. Some chatbots benefit from having access to services that already aggregate information on businesses. Google’s Gemini, for example, can give a list of restaurants when prompted complete with details like hours of operation, menus, and maps because it can pull from Google Maps.

Perplexity plans more integrations like this in the future, and it already works with the answer engine WolframAlpha for mathematical computations. Srinivas provided a few potential examples, like working providers of shopping data or financial data.