There is a quiet moment before every real shift. No fireworks. No drumroll. Just builders staring at a problem long enough to get annoyed, then obsessed, then precise. That is where Linkup lives. Not in the noise of search results made for human eyeballs, but in the narrow lane where machines actually need answers. February 3, 2026 marked one of those moments, with Linkup closing a $10M seed round to build web search for AI, and doing it without pretending the internet was designed for anything other than people clicking blue links.
Philippe Mizrahi (CEO) saw that gap early. The web talks in paragraphs. AI listens in facts. At Lyft, working on Amundsen, Philippe Mizrahi learned that discovery only matters when the system understands what it is looking for. Denis Charrier (CTO) learned the same lesson the hard way, building vector search at Niland before Spotify, where scale teaches humility fast. Boris Toledano (COO) brings the operational discipline, shaped at McKinsey, that keeps ambition from drifting into mythology.
Linkup does not index pages. It breaks the web into atoms of information, tagged, timestamped, sourced, and ready to be consumed by machines that do not care about layout, SEO tricks, or vibes. This is search that links up answers, not links. Structured, real time, and legal by design, with licensed content flowing directly from publishers instead of scraped in the dark like a bad habit nobody wants to admit anymore.
That architecture is why Gradient led the round, with Darian Shirazi leaning in alongside Elaia, Leblon Capital, and Weekend Fund. It is why Seedcamp, Axeleo Capital, Motier Ventures, and OPRTRS CLUB came back. It is why operators like Olivier Pomel, Arthur Mensch, Alex Bouaziz, Shuo Wang, and Florian Douetteau put their names behind it. Builders recognize builders. Infrastructure people recognize infrastructure.
Linkup is already serving thousands of users globally, powering teams like Artisan and KPMG, and rolling out Linkup fast for sub second search when latency actually matters. Deep search when you need context. Fast search when you need truth right now. No scraping. No guesswork. Just information shaped for the way AI actually thinks, not the way humans learned to browse in 2004.
The lesson here is not about raising capital. It is about earning conviction. This team did not chase novelty. They chased alignment between how knowledge exists and how intelligence consumes it. That is why this round happened. Not because the story sounded good, but because the system works.
