The first search engine wars were won over quality and speed. Google’s PageRank algorithm ordered search results with better accuracy than other methods and software systems including the Google File System provided higher throughput over inexpensive commodity hardware. There’s now a second wave of search engines built for AI that are helping bring live or scattered context to AI models. Doing so affords users a richer experience with AI models across a broad set of use cases including research and Q&A tasks.

Last year we were fortunate to meet Phil Mizrahi, Denis Charrier, and Boris Toledano who set out to build Linkup under the mission of constructing a world-class web search engine for AI models and agents. Delivering on this product fundamentally means (1) crawling and indexing the open web, (2) embedding the underlying information within pages of the web and (3) exposing interfaces to developers to make sense of this information. The Linkup team set out to innovate on all of the above product components over the past year and today have announced the world’s most accurate sub-second web search API.

Initially we ran our own tests which have now been validated on SimpleQA Verified which show Linkup’s agent search API outpaces peers and competitors significantly. We were very impressed with Linkup’s results and progress given the team only raised a small seed round previously. As agents and models become more pervasive, specific use cases and applications require instant and high quality search results.

Figure: Sub-second Web Search Performance on SimpleQA Verified

Today we’re excited to announce we’ve led Linkup’s $10 million seed round with participation from Elaia, Leblon Capital, Weekend Fund, existing investors Seedcamp, Axeleo Capital, Motier Ventures, OPRTRS CLUB and leading angel investors including Olivier Pomel (co-founder of Datadog), Arthur Mensch (co-founder of Mistral), Shuo Wang (co-founder of Deel), and Florian Douetteau (co-founder of Dataiku). We encourage you to try out their product.