Manifold.

Somewhere in a lab connected to Manifold, a scientist types a question that used to require a quarter of someone's life: pull every patient matching this molecular profile, harmonize the genomics, line up the clinical history, and tell me what is interesting. The answer comes back in minutes.

This is Manifold in 2026: a Boston-area applied-AI company that builds a vertical agent platform for life sciences. Not a chatbot. Not a slide about the future. A working system that pharmaceutical companies, molecular diagnostics firms, biobanks and academic medical centers use to move multimodal biomedical data from raw and scattered to governed and analysis-ready.

The company calls its core an Agent OS - a layer of AI agents trained on the unglamorous specifics of life-sciences data. The pitch is refreshingly unromantic. Science, Manifold argues, is rarely the bottleneck anymore. The plumbing is.

Modern biomedicine generates spectacular data: genomic sequences, multi-omics panels, imaging, real-world clinical records, registries. The problem is that each of these arrives in its own format, with its own governance rules, and its own gatekeepers. Bringing them together is less like research and more like diplomacy.

So the people paid to think about cancer spend their days reformatting files, chasing access permissions, and rebuilding the same cohort for the fourth time. It is, to put it gently, a poor use of a PhD.

Manifold's founders noticed that the gap between what science could do and what researchers actually got done was widening - not because the ideas were missing, but because the data was trapped. The bottleneck had quietly relocated from the lab bench to the file system.

Translation: the dream is fewer heroic all-nighters reconciling CSV files, more actual discovery.

Manifold began in 2016 as an applied-AI lab, the kind that takes on hard machine-learning problems for other companies. Plenty of firms with that origin story drift toward whatever pays. Manifold went the other way and bet the company on the single most regulated, most fragmented, most consequential vertical it could find: human health.

The founding bench reads like a deliberate mix of deep tech and deep healthcare. CEO Vinay Seth Mohta is an MIT alumnus who came up through MathWorks, Endeca, KAYAK and co-founded Kyruus. Co-founder and President Vivek Mohta brings a research lineage through Harvard Math and MIT Physics. Co-Chief AI Officers Sourav Dey and Rajendra Koppula round out the machine-learning core.

Around them the company stacked veterans from Apple, Allscripts, Kyruus, Apixio, PatientPing and TetraScience - people who already knew that healthcare software fails in expensive, specific ways, and built anyway.

Manifold's platform is less a single app than a connected workbench. Agent OS sits at the center, with specialized agents that understand the shape of genomic, clinical and omics data. Around it sit the tools researchers reach for daily.

Manifold serves tens of thousands of users across hundreds of organizations - molecular diagnostics labs, academic medical centers, real-world data providers, biobanks, omics service providers and population-genomics groups. The Series B math is straightforward: $18M in December 2025, on top of earlier rounds, for roughly $40M total.

Then there is the partnership that tells you who takes Manifold seriously: Foundation Medicine, a heavyweight in molecular diagnostics, brought Manifold's AI-enabled analytics into its FoundationInsights product to speed drug discovery and development.

A guest list where everyone owns a sequencer and nobody enjoys formatting files.

Mission statements usually age like milk. Manifold's is unusually literal: accelerate life-changing medicines to patients. The platform speeds workflows from target identification through clinical development, market access, and precision medicine in the clinic - while holding the governance line that regulated research cannot cross.

That last clause is the hard part. Anyone can move data quickly if they ignore who is allowed to see it. Manifold's bet is that governed speed - access control, lineage, multi-entity collaboration - is the only kind of speed that actually ships in healthcare. The boring features are the moat.

The next stretch is about network effects. Manifold wants companies, teams and AI agents working over a shared, governed data ecosystem - where each new dataset and each new collaborator makes the platform more useful for everyone connected to it. The Series B is pointed squarely at that: more Agent OS, more cross-organization collaboration, more data gravity.

It is a contrarian kind of ambition. While much of the industry chases the flashiest demo, Manifold is automating the part nobody films: the harmonizing, the cataloging, the access requests. The unglamorous middle of medicine.

Return, then, to that researcher and the question typed before coffee. A few years ago it was a project. Today it is a query. Manifold's whole reason to exist is to keep shrinking the distance between asking a question about disease and getting an answer - because, as its CEO keeps pointing out, the time in between is measured in lives.