SAN FRANCISCO (AP) — Online review aggregator Yelp wants to harness artificial intelligence to make it easier for users to find information curated by other people.
Although Yelp’s users have always been able to dive into its reservoir of 330 million local business reviews, they sometimes find themselves drowning in a sea of commentary from other people about restaurants, doctors, plumbers, roofers and a smorgasbord of other merchants — a problem the new chatbot assistant is designed to solve.
For instance, if a user asks Yelp’s new assistant for a good place to get coffee with a dog, the app will show recommendations alongside relevant reviews.
“This chatbot can really understand 500 reviews in a second whereas a consumer might say, ‘Well, I read the first five reviews, so I guess that’s good enough,’” said Yelp CEO Jeremy Stoppelman, who co-founded the company 22 years ago.
Analyzing and explaining vast amounts of information in an easily digestible summary is something that other leading chatbots such as OpenAI’s ChatGPT, Anthropic’s Claude, Perplexity’s answer engine and Google’s AI overviews already have been doing.
Yelp believes its chatbot will stand out by pointing to the reviews that led to its recommendations and conclusions. The San Francisco-based company decided to create an AI chatbot that shows the evidence underlying its findings, after a survey found that most consumers worry the technology provides misinformation or fabrications.
“People want AI chatbots to be transparent about where they are getting the data from, they want to see the reviews alongside the results when they’re doing local search,” said Craig Saldanha, Yelp’s chief product officer. “So we are trying to make sure the human connections stay front and center while AI handles all the drudgery of making those connections.”
Yelp has been looking for a spark during an AI boom that has more than doubled the value of the tech-driven Nasdaq composite index while its stock price is stuck at roughly the same level as the end of 2022, shortly after OpenAI released ChatGPT.
Although Yelp business reviews have always been popular among consumers looking for recommendations about places to eat and shop, the company hasn’t been able to overcome people’s almost reflexive habit of turning to Google when they’re searching for almost anything.
Google was already synonymous with search by the time Stoppelman and Russel Simmons launched Yelp in 2004, but its results about local businesses often were either inadequate or inaccurate.
To help fill the void, Google signed a two-year licensing agreement to gain access to Yelp’s reviews. But the partnership fell apart when Google began to summarize various information about various topics — including restaurant recommendations in a particular neighborhood — in a way that gave consumers less reason to click on links that sent them to other sites.
That phenomenon has hurt Yelp and other free online services that make most of their money from advertising; Yelp depends on Google for more than 70% of its web traffic in the U.S.
The tensions came to head when Yelp accused Google of improperly raiding its business reviews and favoring its own services. Those allegations helped trigger a investigation by the Federal Trade Commission that ended with a 2013 settlement that only required Google to make a few relatively minor changes.
But the complaints about Google’s tactics didn’t stop, triggering a U.S. Justice Department lawsuit that culminated in a 2024 decision condemning the search engine as an illegal monopoly. But a federal judge last year rebuffed the government’s request to break up Google, and instead ordered less drastic changes — a decision that was shaped by the way more people now rely on chatbots to inform them instead of search engines.
Yelp is pursuing its own antitrust lawsuit against Google in a case scheduled for a May 2028 trial.
As part of its effort to diversify and increase its annual revenue of $1.5 billon, Yelp already is licensing some of its data to OpenAI for potential usage in ChatGPT while betting that its chatbot’s emphasis on connecting people will still reel in more traffic to its service.
“With this new technology, we really think you are going to be able to find that needle in a haystack and have a far more personalized experience,” Stoppelman said.
