Drylab AI Raises $3M to Build Autonomous Science Agents

Singapore-based Drylab AI has raised a $3 million seed round led by Vertex Ventures to scale its platform of autonomous research agents built for biopharma and scientific research, with participation from patent-intelligence unicorn PatSnap. The funding round, announced this week, is one of the first venture bets on agents that are designed to conduct scientific work with minimal human supervision, according to the Business Times.

Founded by chief executive John Phan, chief technology officer Huy Phan, and chief science officer Nhi Nguyen, Drylab AI builds agents that can understand a research objective, formulate a strategy, and then select from more than 800 integrated scientific tools and databases to carry out the work. According to The SaaS News, the agents are trained on real scientific analysis runs and learn from methodological decisions and errors made along the way, so they can interpret results and troubleshoot experimental failures autonomously rather than waiting for a scientist to re-prompt them.

Agents that run the experiment, not just answer questions

The distinction Drylab AI is drawing matters for how the agent economy is evolving. Most research assistants today are glorified answer engines: a scientist asks a question, the model retrieves papers or data, and the scientist does the experiment. Drylab AI's pitch, as reported by the Business Times, is that its agents can work independently of constant prompting and supervision, taking an objective and running the full loop of planning, tool use, result interpretation, and error correction.

Co-founder and CEO John Phan said the company's vision is to bring that capability into the daily work of a highly regulated industry like biopharma through a swarm of autonomous agents working alongside researchers, according to the Business Times. That "swarm" framing puts Drylab AI in a growing cohort of startups treating multi-agent systems, rather than single all-knowing models, as the practical architecture for real scientific work.

The platform, which launched in February, already counts paying customers among pharmaceutical companies, cancer centres, biomedical research institutes, and universities across North America, Europe, and Asia, according to the Business Times. Early revenue before a seed round is a meaningful signal in the agent space, where many startups are still running on demos and pilots.

Why compliance money matters in regulated labs

Biopharma is one of the hardest industries for AI to enter, and one of the most valuable if it works. Drug discovery workflows are slow, expensive, and bound by strict data-governance rules, which is why Drylab AI says its platform can be deployed inside a pharmaceutical company's own infrastructure. Keeping data and discoveries on the customer's side of the firewall is a deliberate answer to the industry's oldest objection to AI adoption: nobody wants their next molecule training someone else's model.

The funding will be used to expand the engineering and biology teams, hire sales staff, and qualify for the security and compliance certifications required to serve large research organisations, according to the Business Times. For context, agent startups that succeed in regulated verticals often grow through precisely this playbook — earn the certifications, then use them as a moat against general-purpose assistants. Earlier this week, Phinity Labs raised $5.2M to push autonomous agents into chip design, showing that investor appetite for autonomous specialist agents extends well beyond the life sciences.

A signal about where agent funding is flowing

The Drylab AI round fits a pattern visible across the past month of agent funding. Investors are moving away from generic "AI assistant" startups and toward agents embedded in specific workflows where autonomy has a measurable price tag — order-to-cash, chip design, offensive cybersecurity, and now wet-lab research. The $3 million check is small compared with the nine-figure rounds going to horizontal agent platforms, but seed-stage funding in a vertical agent startup is arguably a stronger signal of where the next wave of enterprise deployment is heading.

There is also a talent geography story here. Singapore's rise as an AI hub has been driven in part by investors like Vertex Ventures, which has deep roots in the region, and by strategic investors like PatSnap, whose patent-intelligence business gives it a natural window into which research directions are heating up. PatSnap's participation suggests the round is as much about domain access as about capital.

The company says the new capital will help accelerate adoption and build locally deployed intelligence that moves scientists from biological questions to validated discoveries faster, according to the Business Times. Whether autonomous research agents can be trusted to run real experiments without constant oversight is the open question the entire sector is now testing. The bet from Vertex Ventures is that in biopharma, the answer will come faster than most people expect.

Drylab AI's progress will be worth watching alongside the broader move toward agents that learn on the job: DigitalOcean's agent learning services are pushing a similar idea in the infrastructure layer, and the convergence of self-improving agents with domain-specific tool access could be the defining product pattern of this agent cycle. For now, the company's seed round is a milestone for a small but growing thesis — that the most valuable agents of the next few years will be specialists, not generalists.

More details on the company are available on Drylab AI's official website.