Nextnet.

Biomedical research has a paperwork problem. Roughly four thousand new papers land on PubMed every day. The pile is not the bottleneck; the connections between the pile are. Which gene talks to which pathway. Which drug failed in which trial, and why. Which two unrelated labs are accidentally working on the same molecule. Nextnet's bet is that artificial intelligence is finally good enough to draw those lines for you - and honest enough to show its receipts.

The product comes in two shapes. Nextnet Copilot answers questions in plain English with citations attached to every claim, so a researcher can click through to the underlying paragraph instead of taking an algorithm at its word. Nextnet Explorer is the more theatrical sibling: an interactive map that turns the literature into a navigable constellation of genes, drugs, diseases, and the human beings studying them. Below both products sits a semantic web stitched together from PubMed, ChEMBL, Ensembl, and friends - the unglamorous plumbing that makes the magic feel like magic.

Steven Banerjee is not a man who picks small problems. His previous company, Mekonos, raised more than forty million dollars trying to make gene editing easier with a chip. Nextnet started in 2021 with an inverse intuition: maybe the hardest part of modern biology isn't the bench - it's the bibliography. So he assembled a small fellowship of co-founders. Jason Collins, a Duke-trained biomedical engineer who had spent years selling B2B SaaS. Malik Alimoekhamedov, a cloud computing specialist with a calm CTO demeanor. Derek Park, an Oxford Marshall Scholar with a doctorate in applied mathematics and a sideline in computational oncology.

The market they entered was crowded with chatbots that promised everything and verified nothing. Nextnet picked the opposite stance. It built the knowledge graph first, then placed the language model on top, then refused to ship a single answer without a citation underneath. The phrase researchers use most often when describing it is "I trust this one." It is not a phrase you hear often in 2026.

Adoption followed a familiar shape for product-led companies: a handful of curious graduate students, then a lab, then a department, then a press logo. Today's customer list reads like a tour of the world's most aggressive teaching hospitals: Harvard, MIT, UCSF, MD Anderson, Columbia, Yale, Berkeley, Emory, Broad Institute. The pharma giant Eisai is in there too. Organic growth hit 1,200% in a year, all without a traditional outbound sales team. That is the kind of number you usually find in slide decks - in this case it appears to be plain arithmetic.

The company is still small. About thirty people, spread across San Francisco and Europe, working remotely. The seed round - roughly 1.3 million dollars from Hike Ventures, ODF, and Propel(x) - is modest by 2026 AI standards. Nextnet is operating on a thesis common to good tools: get the product right, let the users do the selling, raise on revenue when the time comes. So far, the users keep selling.

Fact 01   The head of brand marketing has a PhD in cultural studies and was trained as an art historian. The brand feels like it.

Fact 02   Co-founder Derek Park is a Marshall Scholar - a fellowship designed to send promising Americans to study in the UK, named after the post-war Marshall Plan.

Fact 03   The company tagline - "where researchers love researching" - is the rare tagline that contains a verifiable claim.

Fact 04   Steven Banerjee earned his engineering degree at the University of Canterbury in New Zealand. Then went to Berkeley. Then started two companies.