Generating a Story Portrait for Every Times Subscriber
Enabling In-App Personalization
The New York Times and Droga5 wanted to showcase what it means to be a subscriber in a creative way. They turned to Addition to help them achieve a personalized experience within their mobile app, a textual 'Story Portrait' that’s as unique as every subscriber. By fine-tuning AI on the New York Times’ editorial tone and guidelines, we enabled a personalized experience that adapts to every user based on their reading habits while maintaining the New York Times' distinctive textual and visual style.
As Unique As The Articles They Read
Story Portrait used data from subscribers to help bring their reading history to life. The experience, launched in the NYT App, created a textual representation of every subscriber with the click of a button. By tapping into the Times' data about the articles individuals most engaged with, the experience assembled a unique ‘Story Portrait’ with the help of AI that users could reveal and share on social media platforms.
Finetuned on the Times' Content and Guidelines
To make the campaign possible, we needed to come up with a brand-safe way of generating people’s Story Portraits. To do this, we opted for an approach where we pre-computed over 100,000 pieces of content and used a classical data science algorithm to display the best combination to the user. By building an AI system and fine-tuning an LLM using thousands of articles, human-written examples, and the New York Times’ editorial guidelines, we were able to deliver content at scale that retained factuality and enable the project to launch on a tight timeline. These assets were then used for important figures, such as Quest Love, and featured in print ads, film, and OOH advertising.
Powering an Integrated Campaign
Story Portrait became the focal point of an integrated campaign from the New York Times, featuring personalized outputs from iconic New Yorkers in out-of-home and video ads that helped visualize what it means to be a Times subscriber.
2024