The workflow chart on Enjamb Labs' website shows lines connecting Benchling to Veeva to Medidata to SAS, then back again—each arrow representing a handoff where data gets manually exported, reformatted, and reimported by scientists who'd rather be doing actual science. This is how pharmaceuticals get made in 2026: on digital infrastructure that many in the industry consider outdated.
Rayan Mubarak and Maadhav Deekshitha think they can blow up that map entirely.
Their company, a two-person operation that emerged from Y Combinator's Spring 2026 cohort, has a pitch that borders on the preposterous: collapse every stage of drug development—from initial discovery through FDA submission—into a single browser workspace orchestrated by AI agents. Even more audacious? They report having landed 500 users at some of the world's largest pharmaceutical companies within weeks of launching, though this figure is company-reported and not independently verified.
By June, logos from Johnson & Johnson, AbbVie, Merck, Sanofi, and Bristol Myers Squibb were displayed prominently on Enjamb's landing page, though the specific nature of these relationships has not been independently verified. Whether those represent serious deployments or polite pilot programs, the founders won't say. But the names alone suggest they've managed to get someone's attention in an industry not known for embracing unproven software from startups that could fit in a studio apartment.
$650,000 and a Theory of Connective Tissue
The company announced a $650,000 pre-seed round in April 2026, backed by Y Combinator and Founders Inc. Modest by Silicon Valley standards—barely enough to cover a year's payroll for a small team in San Francisco. But Mubarak and Deekshitha aren't building a conventional SaaS product. They're attempting something stranger: what they call an "agentic workspace," where specialized AI agents don't just assist researchers but actively coordinate the entire research lifecycle.
This isn't another chatbot that answers questions about protocols or summarizes literature. (Those already exist—Elicit and SciSpace have carved out niches there.) Enjamb is positioning itself as connective tissue, a layer that sits on top of pharma's existing infrastructure and makes it all talk to each other. The founders describe "MCP-style connections"—a reference to model context protocol integrations—that link disparate systems without forcing companies to migrate their data.
Whether that's technically feasible at scale, or just optimistic architecture diagrams, remains to be seen.
From Lab Bench to Regulatory Binder
The platform's stated scope is genuinely ambitious. It claims to handle clinical evidence synthesis during discovery, trial design and protocol scaffolding, statistical programming (including the alphabet soup of SDTM, ADaM, and table-listing-figure generation), and regulatory document drafting. All of it supposedly happens in a browser workspace that includes full Microsoft Office editors plus sandboxed Python and R environments where code can run without escaping into production systems.
Enjamb markets "100+ scientific and enterprise integrations" and "50+ task-tuned models for in-silico validation," though the specifics remain vague. Which models? Whose APIs? The company hasn't disclosed details that competitors or skeptical enterprise buyers would want to know.
What it does emphasize is auditability. Every source citation, every tool call, every line of code executed, every document revision—all of it logged and traceable. In an industry where a single data integrity lapse can torpedo a billion-dollar filing, that audit trail isn't a nice-to-have. It's the entire ballgame.
The performance claims, though, are where things get interesting. Or suspicious, depending on your priors.
Enjamb asserts it can generate a complete FDA submission package in 48 hours instead of eight months. It claims "23x fewer errors than GPT-5.5" and says "98% of datasets pass Pinnacle 21 on first pass." For those outside pharma, Pinnacle 21—owned by Certara—is the validation engine that both the FDA and Japan's PMDA use to review clinical data submissions. Getting datasets to pass on the first try is genuinely difficult. Getting them to pass 98% of the time would be remarkable.
If true.
These metrics appear on the company's Y Combinator profile and website as of late June, and as noted later in this article, they lack third-party validation. No case studies, no customer testimonials, no independent audits. Just numbers on a landing page from a startup that's been live for maybe ten weeks.
The Builders
Mubarak comes from biology and machine learning. According to Enjamb's Y Combinator profile, he's first author on a cancer ML paper published in JMIR that improved on state-of-the-art performance by 20%, and he's presented research at the Society of Toxicology and the International Society for Computational Biology. Not a typical startup founder résumé, perhaps more than you'd expect from someone launching consumer software—but exactly the background you'd want if you're trying to automate oncology trials.
Deekshitha was the youngest engineer at Dell's AI Lab, working on AI CPU optimization for supercomputing infrastructure, and previously held an R&D role at Broadcom. The kind of person who thinks about compute at scale before most engineers have their second cup of coffee.
Together, they're still just two people. Y Combinator's Spring 2026 directory lists the headcount at two, though LinkedIn suggests "2-10 employees"—startup speak for "we might have some contractors, but we're not saying." By late spring, Mubarak was posting on LinkedIn about the philosophical difference between Q&A-focused language models and "execution agents for running drug programs," and turning up at industry conferences like SynBioBeta 2026.
For a team this small to be claiming users across multiple Fortune 500 pharma companies feels... ambitious. Or perhaps naive. Time will clarify which.
The Freemium Gambit
Enjamb's pricing starts at zero. A free tier with 1GB of storage, designed to let individual researchers kick the tires. From there it scales: $20/month for Pro (billed annually), $40/month for Max, and $40 per seat monthly for the Lab tier aimed at teams. Enterprise pricing is custom, naturally, with the usual enterprise features—single sign-on, SAML integration, "unlimited usage."
The strategy is familiar to anyone who's watched developer tools take over enterprises from the bottom up. Get individual contributors hooked, let them evangelize internally, watch IT departments eventually capitulate and cut a check. Slack did it. GitHub did it. Notion, Figma, the list goes on.
But pharma isn't software. The compliance requirements alone—21 CFR Part 11 for electronic records, ICH E6(R2) for good clinical practice, GDPR if you're touching European patient data—make bottom-up adoption harder. A researcher can't just swipe a credit card and start running clinical trials through a new platform, no matter how slick the UX.
What's Already Out There
Enjamb is hardly entering virgin territory. Formation Bio, for instance, raised significant capital to build clinical trial infrastructure, with its CEO telling TIME in February 2026 that "the biggest problem in the pharmaceutical industry hasn't been drug discovery... [it's] running clinical trials." Medidata announced that same month it had supported more than 500 clinical studies with AI over the past decade—not exactly a startup scrapping its way to product-market fit.
In May, Applied Clinical Trials reported that early AI adopters were seeing reduced timelines, with priorities shifting toward predictive trial design. The market is clearly moving. The question is whether there's room for a two-person startup to wedge itself between established systems-of-record like Benchling and Labguru (which own the data infrastructure) and specialized AI assistants focused on narrower use cases.
A June analysis from Fluenta, which scored Y Combinator's Spring 2026 batch, flagged exactly those competitive pressures. The report questioned Enjamb's focus and whether it could overcome switching costs in enterprise pharma—a reasonable concern for any startup trying to displace entrenched vendors in a risk-averse industry.
Yet the user numbers, if accurate, suggest something's resonating. Five hundred users across major pharma companies in three weeks isn't nothing, even if "user" might mean "downloaded the browser extension and logged in twice."
The Regulatory Tightrope
Here's where it gets genuinely tricky.
Pharma operates under FDA oversight that demands near-perfect documentation. Every decision, every data point, every statistical method must be defensible under regulatory scrutiny. Enjamb is promising to automate enormous chunks of that documentation—protocols, statistical analysis plans, investigational new drug applications, new drug applications, entire briefing books—while maintaining the audit trails regulators require.
If the Pinnacle 21 claims hold up under real-world conditions, that matters. A lot. But drug development failures rarely come from obvious mistakes that automated QC can catch. They come from subtle errors compounded across thousands of judgment calls, from contextual nuances that don't show up in a dataset but matter enormously to a medical officer reviewing your submission.
Can AI agents navigate that? Can they handle the soft knowledge—institutional memory about what specific FDA reviewers care about, the unwritten norms of how certain therapeutic areas expect to see data presented—that currently lives in the heads of expensive regulatory consultants?
Maybe. The technology is moving fast enough that dismissing it out of hand seems unwise.
But asking a two-person startup to reliably handle workflows that typically require teams of PhDs, statisticians, and regulatory specialists—with billions of dollars riding on getting every detail exactly right—is a different proposition than building a better note-taking app.
An Uncertain Trajectory
For now, Enjamb has a product, user claims that range from intriguing to implausible, and the imprimatur of Y Combinator. What it doesn't have is a track record. No published case studies showing a drug that went from discovery to FDA submission on their platform. No public customer willing to go on the record about how the software performed when regulators started asking hard questions.
The founders are making an aggressive entry into one of healthcare's most conservative sectors with software that promises to fundamentally reshape how drug development happens. That's either visionary or reckless, and which one depends entirely on whether the technology can deliver what the marketing promises.
The pharmaceutical industry has been waiting decades for someone to untangle its digital infrastructure. Whether two engineers with $650,000 and some AI models can actually pull it off—well, that's the kind of question that only gets answered in the field, under regulatory scrutiny, with real drugs and real patients in the balance.
