Humor Genome maps the DNA of humor: an open taxonomy, dataset, and set of benchmarks to help AI understand what is funny across cultures, formats, and contexts.
Humor Genome is research driven, but it stays close to real performance. It powers tools that measure laughter, compare interpretations, and simulate audience response before live testing.
Align audience laughter to transcript lines, visualize joke impact, and generate data grounded feedback to improve a set.
Multi agent interpretations compared via Elasticsearch ES|QL, helping creators see how work may be understood before live testing.
Workshop a set in real time with four specialized voice agents powered by Gemini 3 and 2.5.
I am looking for collaborators across research, engineering, and creative practice. If you care about modeling interpretation, measuring audience response, or building tools that help creators iterate faster, reach out.