Perplexity CEO Aravind Srinivas speaks The Verge Many people use chatbots just like regular search engines. It makes sense to provide users with the information they are looking for, such as restaurants, directly from the source. So when people ask for restaurant recommendations, we integrate Yelp maps, reviews, and other details in response.
“Our fundamental motivation is that we started as a tool to appeal to fact-checking and research people. But research can be anything. For example, if you're a coffee addict, where are good coffee shops? You want to know,” says Srinivas.
as The VergeDavid Pierce recently pointed out that how Perplexity combines text and links when answering questions is already a selling point compared to other chatbots such as Copilot, Gemini, and ChatGPT. did. Srinivas said that consideration is part of why the company ultimately chose his Yelp. Srinivas said Yelp's data allows Perplexity to provide answers in a variety of ways, including adding maps, photos, and even citing reviews.
The response includes a link to the business's Yelp page, so users can see more reviews and other details. Perplexity did not disclose financial details of the deal and said it does not use Yelp data to train its models, as the company uses existing models such as GPT and Claude 2 for its chatbots.
This connection is part of Yelp's Fusion program license data for third parties, including BMW. “By integrating our rich, highly trusted local data into Perplexity, we will be able to provide our users with reliable local search results,” said Chad Richard, SVP of Business and Corporate Development at Yelp. said in an email: The Verge.
Perplexity is the latest AI company to contract with a continuously updated source of up-to-date information, but Srinivas said the company is not affiliated with Yelp, but licenses its data. points out. Google will reportedly pay Reddit $60 million annually to access real-time data and use Reddit data to train AI models. OpenAI has signed deals with news publishers like Axel Springer to obtain data for training.
Other chatbots can also suggest cafes and restaurants, but much of this data comes from information models learned from the internet, as well as user searches. Some chatbots benefit from having access to services that already aggregate information about your business. For example, Google's Gemini comes from Google Maps, so you can enter details like opening hours, menus, and maps to see a list of restaurants.
Perplexity plans more such integrations in the future and is already working with WolframAlpha, an answer engine for mathematical calculations. Srinivas gave some potential examples, including real-world providers of shopping and financial data.