Basecamp Research’s Metagenomic Foundation Model Now in Claude Science

Biological progress in the era of computation, and now AI, has been constrained not by progress in the underlying science, but equally by structural friction: fragmented toolchains, siloed and narrow data sets, and the significant temporal latency between hypothesis formulation and empirical validation.  

Basecamp Research was prescient in seeing that early. In "Biology's Bitter Lesson," the team noted that foundation models trained on the full breadth of biological information encoded by nature over billions of years will drive significant progress toward programmable medicine. 

Today, more of that vision comes to fruition as the company's foundation model for antibiotic design and vaccine target prediction, EDEN, becomes available in Anthropic’s Claude Science, delivering the largest metagenomic model in biology to every scientist’s desktop.

For scientists and researchers around the world, this opens new, meaningful avenues of working. Designing multiple drug candidates, reasoning them against the deepest map of natural biology ever assembled, and working through the results can now happen at an accelerated pace.

The new foundation model 

A foundational model is only as good as the scale and diversity of biological information underneath it. Most biological models learn from the same public databases, which means they share the same blind spots and are already narrowed down due to specialist models and expert heuristics. 

Basecamp built something different. EDEN, which is trained on proprietary genomic data more than 20x larger than any public resource, learns the language of DNA from over a million newly-discovered species. 

That depth is what lets it reason across biology that the field hasn't fully mapped, and design new therapies for cancer and genetic disease directly from a disease prompt. The level of general intelligence required in biology clearly requires orders of magnitude more information than available today to researchers and drug scientists in the field.

From a tent on ice 

Glen Gowers and Oliver Vince met on the first day of their degree at Oxford in a bar, along with Phil Lorenz, now the company's CTO. Twelve years later, that's the founding team.

What drew us to them was an instinct to keep going deeper and to understand the world not just as we see it, but at the molecular level. That curiosity took them somewhere unusual. Retracing a 1932 expedition across the Vatnajökull icecap in Iceland, they compacted a full DNA sequencing lab into portable gear and ran it in the wild — the first off-the-grid DNA sequencing.

More than half of what they sequenced had never been seen before. They came to call it the dark matter of biology. The deeper insight was that public DNA data had been stripped of its context — and that where something lives has everything to do with why it's there and what else evolved to live alongside it. Capture that context across every biome on the planet, and you change how proteins, which are both therapeutic agents and targets, are discovered and designed.

The name itself tells you something about how they think. "Basecamp" came from the 1932 explorers' map, where the word marked the team's home base, and it doubles as a nod to the base pairs in DNA. It's a small detail, but it captures the founders' intent to connect the exploration from those expeditions to the science they helped improve, bridging the physical world to the molecular one. 

What struck us in every conversation was that Glen and Oliver weren't chasing a single discovery. They saw that the way biology had been cataloged for decades was one-dimensional and biased by a narrowing down vs. learning from all biology. 

Their ambition was to rebuild that foundation entirely: a living, interconnected map of life across organisms, rich enough to inform new therapeutics, diagnostics, and materials that don't exist yet. That scale of ambition rooted in deep science and insight is exactly why we back this type of founder early. 

That's the bet we are designed to recognize and execute at True Ventures. And it's why this milestone where Basecamp accelerates through collaboration and partnership with other AI leaders feels so significant. 

Progress toward general intelligence in bio

The most powerful biological model in the world only changes science if it's where scientists actually work, with ground truth biological information learned from multiple species, and enabling computational access that permits AI agents and researchers to work effectively. Removing that friction, and embedding this capability across AI environments like Claude Science, makes the question "What should I design” possible and practical. In addition to Claude Science, this new capability is now available in Claude.ai, Claude Desktop, Claude Mobile, Claude Code, and Claude Cowork through Anthropic's connectors directory.

That subtle shift, multiplied across thousands of researchers and computational insights, is how the pace of biological discovery dramatically changes for the better.

We believed in Basecamp early, when the scale of the ambition was still a vision in a tent on a glacier. Today, a great deal more of the world gets to see why, and this is only the beginning of a journey that will continue their march toward a trillion genes and fundamentally new biological insights in favor of better human health for everyone.

More on the news here.