Yelp has begun rolling out artificial intelligence-generated videos across its platform, aiming to enhance user engagement by creating short, visually engaging content about restaurants, food spots, and nightlife businesses. These AI-stitched videos, now widely available on Yelp’s iOS app, form part of a TikTok-style vertical feed on the app’s homepage.
The initiative marks Yelp’s latest use of generative AI to curate dynamic content from user-submitted photos, videos, and reviews. After a period of limited testing that began in 2023, the company has now extended the feature to all iOS users in the United States.
How the AI videos are made
Yelp’s AI-generated videos are crafted through a collaboration of several artificial intelligence tools. The process begins with OpenAI’s large language models, which analyse content to write descriptive scripts and select storytelling angles. These scripts are then narrated using ElevenLabs’ AI voice technology, while Amazon Transcribe generates on-screen captions that match the spoken content.
The resulting videos showcase highlights such as food quality, drink offerings, atmosphere, and overall customer experience. Each video combines a selection of images and clips uploaded by users, stitched together to form a seamless visual narrative with voiceover and captions.
Although these videos are created from publicly available user content, individuals currently cannot opt out of having their media used in this manner. Additionally, business owners are not given previews or controls over the videos generated for their listings. Only one AI-generated video per business is published at a time, and it may be updated over time if new content becomes available.
Content moderation and personalisation
Yelp has stated that AI-stitched videos are only produced for businesses that have sufficient visual and textual material to tell a compelling story. “We want to make as many videos as possible,” said Craig Saldanha, Yelp’s Chief Product Officer, in a recent interview. However, Saldanha clarified that the creation of these videos depends heavily on the volume and quality of a business’s existing photos, reviews, and videos.
The decision to show a video to a specific user is based on personalised engagement signals, such as previous activity and preferences. However, the videos themselves are not customised per viewer—there is only one standardised video available for each business at any given time.
Suppose a video is considered inaccurate or inappropriate by either a user or a business. In that case, Yelp provides a reporting mechanism accessible through the three-dot menu located in the top right corner of the video interface. According to Saldanha, Yelp also conducts routine large-scale audits to ensure the accuracy and appropriateness of the AI-generated content.
Part of a broader AI strategy
The new AI-generated videos are part of Yelp’s broader efforts to incorporate artificial intelligence into its platform. Recent additions include AI-powered review summaries and intelligent filters that help users navigate large volumes of reviews more effectively.
By leveraging generative AI, Yelp aims to provide users with a more engaging and visually appealing way to explore local businesses, while also enhancing the discoverability of venues with strong user-generated content. The company appears to be committed to expanding these tools, as it continues to explore ways to use AI to improve user experience and business visibility.
