The most expensive object in pharmaceutical market research may be a PowerPoint nobody can find. A brand team spends six figures on a study, circulates the deck, makes a decision and moves on. Months later, a slightly different question appears. The old evidence sits in a folder. The institutional memory has changed jobs. Another brief goes out.
ZoomRx has built its current company around this awkward loop. Founded in 2009 by Sriram Subramanian and Karthik Kalyanam, the Cambridge, Massachusetts firm began as a specialist in healthcare research and analytics. It recruited healthcare professionals for mobile surveys, analyzed promotional performance and helped drug companies understand what physicians and patients were actually doing. Today it presents itself as something broader: a vertically integrated life-sciences intelligence business with consulting, proprietary panels, commercial analytics and two AI platforms, Sagan and Ferma.
The buyer is not a patient looking for medical advice. It is the machinery behind a drug: brand leaders preparing a launch, insights teams testing messages, medical-affairs groups watching a congress, commercial analysts allocating sales representatives, and executives trying to understand why a promising therapy is not landing. ZoomRx says it works with more than 100 biopharma companies, including 19 of the global top 20. Its public client display includes familiar names such as Johnson & Johnson, AstraZeneca, BioNTech, Gilead, Amgen and Regeneron.
01 / The old machineWhat failed first was the handoff
There is no public founder confession about a disastrous first product, and inventing one would make a neat story dishonest. The visible failure is more ordinary: the handoff between evidence and action. Claims data can show what happened but rarely why. A survey can capture attitudes but miss the exam-room language that shaped a prescription. A competitive-intelligence database can surface an event but leave an analyst to turn it into a briefing. A consulting study can answer one question while forgetting every question that came before.
ZoomRx attacked those breaks one by one. Promotional Effectiveness Tracking measures what healthcare professionals remember after sales interactions. Awareness, Trial & Usage studies track how a brand moves from recognition to prescribing. Patient Chart Audits and HCP-Patient Conversations reach closer to clinical reasoning. Decision Sciences combines primary research with claims, prescription, CRM and sales data. The 2024 acquisition of Vivisum Partners added Brand Builder, a program designed to shape a therapy from early clinical development through launch.
“Enterprise AI in biopharma lives or dies on trust.”Sivakumar K, VP Technology, ZoomRx
That acquisition also reveals where ZoomRx changed its posture. Historically strong in post-launch tracking, it moved upstream into pre-launch strategy. Then its product releases moved in the other direction - beyond a finished engagement and toward continuous infrastructure. This is an inference from the sequence, not a peek inside a board meeting: acquire the early-stage work, connect the research archive, monitor the outside market, and remain useful across the full commercial life of a drug.
02 / The product mapTwo planets, one intelligence system
Ferma watches the world outside the customer. Ferma Congress covers medical meetings, with ZoomRx claiming more than 250 conferences across nine therapeutic areas since 2020. It can summarize sessions, prioritize a planner and answer questions while news is still fresh. Ferma Agents monitors pipelines and competitive sources, then produces memos, landscape models and reports in Word, Excel or PowerPoint. The choice of output is telling. ZoomRx is not asking a pharma committee to fall in love with a chat window. It is putting the answer into the files that committees already review.
Sagan works on the customer’s own world. The first Sagan release in 2025 used AI-facilitated voice and text conversations with real healthcare professionals and patients. Sagan Agents, launched in May 2026, expanded the idea into a continuous-intelligence platform. Its Data Archive makes old quantitative data, transcripts, claims, reports and CRM material queryable. Agentic Market Research can design, field, analyze and report a new study. The result flows back into the archive, so the next question starts with more context.
The newest piece, Signal-IQ, launched in July 2026. It turns de-identified mobile-location signals into weekly or biweekly estimates of competitor sales-force activity around targeted healthcare professionals. ZoomRx says 2026 backtests converged with client call files at rates above 85 percent for tested brands. It is a clever answer to a slow-data problem: the industry fallback, CMS Open Payments, can arrive 6 to 18 months late. It is also the product with the clearest caution label. Mobility-derived intelligence must fit local privacy, health-data and employment rules, and a backtest is not the same as independent validation.
03 / The economicsConsulting revenue meets software memory
ZoomRx is private and does not publish audited financials or a valuation. A supplied company record estimates annual revenue at $64.3 million and headcount around 550; LinkedIn shows roughly the same employee scale. The business model is easier to see than the accounts. The company sells custom research and strategic engagements, recurring tracking programs, analytics work and enterprise software. Sagan Agents uses customized annual pricing by brand and module. At launch, ZoomRx described the price as roughly comparable to one large traditional market-research study, while offering some existing customers a free archive pilot.
That price comparison is the pitch in miniature. A traditional study is a purchase. A connected archive is supposed to become an asset. If an insights team can recover an old answer, build a cross-study analysis in minutes and field only the missing research, the software may avoid duplicated work. If the archive is thin, messy or rarely consulted, the same platform becomes an expensive search box. Continuity has value only when there is something worth continuing.
04 / The moatThe AI is not the scarce part
A generic model can summarize a transcript. It cannot casually reproduce 17 years of therapeutic context, a verified respondent network or the trust needed to record real clinical conversations. ZoomRx reports a 60,000-plus HCP and patient panel, a 99.8 percent sample-fulfillment rate and 700 million-plus benchmark data points. Those numbers are company claims, but they describe the right kind of defensibility: supply, history and workflow access.
Competitors come from every side. Kantar and Ipsos bring global research scale. IQVIA owns formidable healthcare data assets. ZS Associates and Inizio offer broad commercialization expertise. Cortellis, GlobalData, Evaluate Pharma and AlphaSense occupy parts of the intelligence workflow. Verilogue specializes in physician-patient conversations. ZoomRx’s answer is vertical integration: own the panel, software and consulting judgment rather than pass a project through several vendors.
The human-in-the-loop language is less futuristic than it sounds, and more useful. Pharma work is reviewed by medical, legal, regulatory and commercial stakeholders. A hallucination with a handsome chart is still a hallucination. Sagan’s evidence trails and role-based access controls aim to make each claim traceable; consultants remain responsible for clinical accuracy and strategic meaning. The AI does pattern recognition and production. The expert decides whether the pattern deserves a budget.
05 / The stealWhat another operator can copy
Own the awkward input
Panels, permissions and messy historical files are harder to clone than a polished chat interface.
Start with one delay
Ferma’s useful unit is not “AI.” It is the hours between a conference signal and a briefing.
Ship into the workflow
A finished PowerPoint may be less glamorous than a dashboard, but it travels through the approval chain.
Let work compound
Feed every engagement back into a governed archive so the next job begins with memory, not a blank page.
There is a final lesson in ZoomRx’s culture language. The company has publicly emphasized quality, velocity, experimentation, respect and collaboration, along with hybrid work and flexible hours. Its more recent values stress being knowledgeable, credible, transformative and ingenious. Those are aspirations, not an employee census; review sites show a mix of enthusiasm and criticism. Still, “quality” beside “velocity” captures the product problem neatly. Faster research that nobody trusts is not progress. Meticulous research that arrives after the launch decision is archaeology.
06 / The limitsWhen the model does not work
ZoomRx is a poor fit for a small company with one occasional research question, no archive, little first-party data and no team to govern the output. A generalist research problem outside life sciences would waste its specialization. A clinical or regulatory decision that requires validated medical evidence cannot be delegated to commercial market intelligence. Signal-IQ should not be used where mobility-derived analysis conflicts with law, policy or reasonable privacy expectations. Synthetic audiences may help pressure-test an idea, but they should not be presented as authentic physician or patient testimony.
The larger risk is organizational. Continuous intelligence only works when teams share data, maintain permissions, challenge weak outputs and change decisions when the evidence changes. Software cannot rescue a company that treats every study as a political artifact. Nor can vertical integration guarantee objectivity; the same vendor collecting, analyzing and advising on data demands strong transparency about methods and limitations.
Yet the bet is coherent. ZoomRx spent its first decade and a half collecting the pieces most new AI companies lack: respondents, benchmarks, therapeutic fluency and buyer relationships. Now it is packaging those pieces as memory. The winning feature is not that Sagan can answer a question. It is that, if the system works as promised, the answer does not disappear when the meeting ends.