ZeroEntropy

AI Search · AI Infrastructure · Enterprise Software

Acquisition

Notion acquires ZeroEntropy, absorbs AI search startup's entire team

The startup's reranker had already cut Notion's search latency 85% while replacing tools from bigger, better-funded rivals.

• Notion announced it has acquired AI search infrastructure startup ZeroEntropy, with the entire team joining and founder/CEO Ghita Houir Alami now leading Notion's newly formed Model Research team.
• Founded less than two years ago by Houir Alami and CTO Nicholas Pipitone, ZeroEntropy built custom kernels, an inference stack, and specialization recipes for high-efficiency task-specific AI models like zembed-1 and zerank-2.
• Notion had already been running its unified AI search on ZeroEntropy's zerank-2 reranker for two months, gaining up to 30% faster search and an 85% cut in reranking latency while lowering inference costs.
• As part of the deal, ZeroEntropy open-sourced all of its models under Apache 2.0 and will sunset its products, including its zsearch search engine, by September 4, 2026, with new signups already disabled.
• ZeroEntropy was a Y Combinator-backed startup that had raised just a $4.2 million Seed round roughly a year before the acquisition, making this a quick outcome for a very young team.
• The deal shows platforms like Notion increasingly acqui-hiring small, specialized AI infrastructure teams outright to undercut costlier, larger AI vendors rather than licensing off-the-shelf models.