Aravind Srinivas, CEO of Perplexity, said that many people now use chatbots in the same way they use conventional search engines. Therefore, it makes sense to provide people with information directly from the source, such as restaurant recommendations. When users ask for restaurant recommendations, Perplexity integrates Yelp's maps, reviews, and other details to provide more accurate and useful responses.
"Our initial positioning was as a tool to attract those who need fact-checking and research. But the scope of research is very broad, such as for those who love coffee, they may want to know which coffee shops are worth visiting," explained Srinivas.
Compared to other chatbots like Copilot, Gemini, or ChatGPT, Perplexity's approach to answering user questions, which combines text and links, has become one of its major selling points. Srinivas stated that this was also one of the reasons they ultimately chose Yelp. By leveraging Yelp's data, Perplexity can answer questions in multiple ways, such as adding maps, photos, and even directly quoting reviews.
In addition, the replies also include links to the business's Yelp page, making it convenient for users to further explore more reviews or other detailed information. Although Perplexity did not disclose the financial details of this deal, they made it clear that they will not use Yelp's data to train any models because they currently use existing models like GPT and Claude 2 to support their chatbots.
This collaboration is part of Yelp Fusion, a program that aims to authorize third parties, including BMW, to use its data. "By integrating our highly trusted and rich local data into Perplexity, we are able to provide more reliable local search results for its users," said Chad Richard, Senior Vice President of Business and Corporate Development at Yelp.
It is worth noting that although Srinivas pointed out that Perplexity did not establish a partnership with Yelp but rather adopted a data authorization approach, Perplexity is the latest AI company to strike a deal with a company that can provide continuously updated and up-to-date information sources. It has been reported that Google pays Reddit up to $60 million annually to access real-time data and use it to train its AI models. Similarly, OpenAI has also signed agreements with news publishers such as Axel Springer to obtain the necessary training data.
Of course, other chatbots can also recommend coffee shops and restaurants, but most of this data comes from information models learned from the internet, and their operation is almost like conducting a search for users. However, some chatbots benefit from being able to access services that have already aggregated business information. For example, Google's Gemini can provide a list of restaurants with detailed information such as opening hours, menus, and maps when asked, as it can extract relevant information from Google Maps.
Perplexity plans to carry out more such integrations and has already partnered with the answer engine WolframAlpha to collaborate on mathematical calculations. Srinivas also mentioned some potential examples, such as partnering with companies that provide shopping or financial data.
