[P] Lattice – Track what top AI labs are publishing daily across 24 research topics

A new, free tool called Lattice has launched to help professionals cut through the noise of daily AI research. Built by a developer to solve their own tracking problem, Lattice automatically ingests papers from sources like the Semantic Scholar API, arXiv, and HuggingFace Daily Papers. It then uses a sophisticated pipeline to match authors against a database of over 500 research organizations, including top labs like OpenAI, Anthropic, DeepMind, Meta FAIR, and KAIST.

The platform organizes research into 24 topics across four key categories: Safety & Alignment, Capabilities, Infrastructure, and Applications. Its core feature is a 'Research Radar' that visualizes which topics are accelerating week-over-week, providing a macro view of the field's direction. For each of the 20 tracked labs, Lattice creates profiles showing output velocity and a breakdown of their topic focus. Users can also receive a weekly digest highlighting the most-read papers and breakout topics.

Technically, the stack is built on Next.js 16 with Supabase (Postgres) and deployed on Vercel. A GitHub Actions cron job runs daily at 6am UTC to fetch and process new papers. The summarization is handled by Google's Gemini Flash model. A key design choice is that no login is required; the tool is fully anonymous, storing user preferences locally. This lowers the barrier to entry for researchers, engineers, and investors who need to stay current but are overwhelmed by the volume of publications. By automating discovery and summarization, Lattice shifts the user's effort from finding information to synthesizing it.