NLGJA 2026 National Convention

We’ve spent the last two years figuring out the best way to use AI for investigative reporting in ways that are scalable to the newsroom and easy-to-use for non technical reporters. This hand-drawn presentation will walk the audience through the design and testing of Cheatsheet, an AI tool built by The New York Times for investigative reporting. Instead of a "black box" chat interface, it is centered on prioritizing the primary sources. The session will explore the internal user-testing process, explaining why the Times chose to build custom tooling from scratch rather than relying on off-the-shelf products. This session is designed for investigative reporters, designers, and developers who want to see how the Times bridge-builds between editorial needs and technical implementation. Attendees will leave with a practical roadmap for building AI tools that prioritize the goals of investigative reporting (in-depth understanding, needle in a haystack, classification and mass translation of large amounts of documents), a list of "Dos and Don’ts" for AI in the newsroom, a demo of the tool, and a first look at our plans for open-sourcing Cheatsheet for broader access.