Haaretz participated in the AI Launchpad programme, run in partnership by the Google News Initiative (GNI) and FT Strategies. AI Launchpad was a five-month long programme, aimed at helping publishers deploy AI experiments responsibly and increasing AI literacy in the EMEA news ecosystem.
Over the course of the programme, AI Launchpad helped publishers seize the opportunity that AI can offer by identifying and testing the AI use cases that can make the biggest impact, starting with a problem and adopting a business-first approach.
A leading publication in Israel, Haaretz is known for its high-quality journalism and influential cultural commentary. As a subscription-first business, Haaretz has cultivated a loyal and discerning readership, both in Hebrew and through its English-language edition.
One of Haaretz’s core strengths is its cultural coverage, particularly in film, television, and columns by well-known writers. Recognising an opportunity, Haaretz chose to develop an AI-powered chatbot that could offer personalised movie and TV recommendations based on its editorial archive of culture-focused content.
Designed in-house with support from FT Strategies, the tool enables users to receive tailored movie and TV recommendations based on genre and platform, from Haaretz's archive of 2,400 metadata-tagged reviews. The chatbot was built using Gemini, selected for its Hebrew-language capabilities after other LLMs had proved inconsistent in tests.
While the tool is in early testing, initial learnings highlight some positive initial outcomes. Firstly, the experiment was the first time Haaretz’s data science and product teams had worked closely together on a shared project, setting a precedent for future experimentation within the organisation.
In addition, the process of preparing content for the chatbot significantly improved understanding of the role of data structuring in producing effective AI tools. Thoughtful prompt design and topic restriction ensured the chatbot aligned with the publication’s values and avoided reputational risks.
Haaretz’s hypothesis is that the chatbot will drive stronger reader interaction with its cultural archive, increase session duration, and create a more engaging digital experience for subscribers.
