Clinical Trials AI
K

Komodo Health

Komodo Health sells real world data and AI analytics to drug makers, payers, providers and the consultants and investors who serve them. Its Healthcare Map links more than 1 trillion records on over 330 million deidentified US patients across claims and other sources, refreshed daily, and teams use it to study treatment patterns, size populations, assess trial feasibility, build external comparator cohorts and plan launches.

Marmot, launched in August 2025, is the AI layer. Navigator turns a research question into an analytical workflow, Cohort Insights builds patient definitions and reports a user can review, and Builder lets analysts open, edit and rerun the SQL, Python or R behind an answer. Saved methods and definitions carry forward with version history, and the language models come from Anthropic and OpenAI. Novo Nordisk says studies that took six months now take as little as a weekend, and by November 2026 Komodo counted more than 90 accounts on Marmot. MapLab Enterprise, MapEnhance specialty data and custom analytics round out the portfolio.

Customers include Merck, Alnylam, Novo Nordisk, argenx and Zevra, and Komodo is a CMS Qualified Entity. Arif Nathoo and Web Sun founded it in 2014; it is based in New York with offices in San Francisco, Chicago and Chennai, and has raised about $514 million. Access runs through enterprise agreements.

AI Health Index verifiedNovember 19, 2026
Compare Komodo Health with other vendors
Founded
2014
Headquarters
New York, New York
Categories
clinical-trials-ai
Indexed Products
Marmot Navigator, Marmot Cohort Insights, Marmot Builder, Healthcare Map, MapLab Enterprise, MapEnhance
Buyer Segments
Life Sciences, Pharma, Health Plan, Health System
Assessment

Capability Axes

An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read

AI Capability
BB on AI CentralityThe model is the engine of a core module. The platform carries other value, but this capability does not exist without it.
Vendor Published

Marmot is the AI engine across Komodo's analytics. It turns a plain language question into cohort definitions, queries, scripts and a written report, working from the Healthcare Map and a knowledge graph Komodo built from millions of earlier analyses. Navigator, Cohort Insights and Builder give business users and technical analysts different ways into the same engine.

Take the AI away and conversational analysis and code generation go with it, but the data asset and Komodo's conventional analytics, MapLab and data enrichment products keep their value. The company combines a core AI capability with a large data and software business.

BB on Autonomy and Oversight ModelThe oversight structure is described and one part is missing, commonly the threshold at which the system stops or what happens after it is wrong.
Vendor Published

Analysts can read the code Marmot generates, edit cohort definitions before anything runs and rerun work with version history. An evaluation gate checks outputs before delivery, and each analytical step is logged.

Those controls sit beside a homepage heading that promises the answer is right every time you ask. Reproducible code can reproduce a wrong assumption, and an automated gate does not make every answer correct, so the claim runs ahead of the controls. How much runs automatically, what thresholds trigger escalation and what happens when validation fails are not fully described.

BB on Model and Technology TransparencyThe approach or the suppliers are named without the version and update discipline behind them. Naming a supplier is the entry to this band both here and on Model Supply Chain Disclosure, which ask different questions of the same disclosure: who receives the data, and what produces the output.
Vendor Published

Language models plan and carry out each analysis, surrounded by a semantic layer, reusable methods and specialized analytical tools, and Anthropic and OpenAI are named as the model suppliers. Marmot writes SQL, and through Builder works in Python and R, saving each version. A knowledge graph built from millions of prior analyses supplies healthcare context.

The exact model versions, routing rules, any adaptation data and the replacement schedule are not given. Saved code and intermediate files make a single analysis inspectable, but they do not record which model release produced each step.

BB on Model Supply Chain DisclosureSubstantial partial disclosure, or a chain that is structurally short: an in house build, a cleared model that cannot be quietly swapped, or a deployment where the transfer does not occur at all. Naming the model provider exits the band below into this one; the axis rises from here on the completeness of the party list and on the terms that govern data once it arrives.
Vendor Published

Anthropic and OpenAI supply the language models, Amazon provides cloud hosting, Databricks supports analytics, Snowflake handles warehousing and Datavant supplies tokenization and privacy technology. The subprocessor inventory lists processing countries and separates providers used only by the Mavens business from the rest of Komodo.

That is a named supply chain. What each model provider keeps of a prompt, what training rights it holds and which supplier handles which Marmot workflow are not set out, and the commitment that customer data does not train Marmot has no separate schedule for the outside model providers.

CC on Clinical and Operational EvidenceNamed customers, or vendor reported percentages with no method, denominator or reference standard. Scale of use is recorded here and is not treated as evidence of benefit.
Vendor Published

Merck, Alnylam and Novo Nordisk are named Marmot users. Novo Nordisk's head of clinical data science says studies that used to take six months now take as little as a weekend, without saying how the work compared. By November 2026 Komodo reported more than 90 accounts and over 50 use cases.

That is adoption and customer reported speed, not a controlled measure of benefit. A benchmark of Marmot's accuracy sits behind a contact form, so its question set, scoring and independent review are not visible, and published research that used the Healthcare Map validates the data rather than Marmot's generated analyses.

BB on AI Safety and PHI StewardshipCategorical commitments are published, such as no training on customer data, without the retention schedule behind them.
Vendor Published

The Healthcare Map is built from deidentified patient journeys, tokenized through Datavant, with a commitment never to reidentify them. Marmot states that customer data does not train its models and that each customer's environment is isolated.

The general master agreement, though, allows Customer Data to be used for product improvement and defines model intellectual property broadly, and how those terms fit Marmot's promise is not explained. After termination, customers get 30 days to retrieve data rather than a fixed deletion date. Query logs, source traces and output evaluation help investigate a bad analysis, but no model incident procedure or prompt retention schedule is described.

Regulatory and Compliance
CC on HIPAA and BAA PostureA status is stated but not supported: compliance is claimed, or business associate status stated, without the underlying document, including where the only privacy notice published covers the website rather than the service that handles patients. The mirror case grades here as well: a substantive privacy document that reaches the service, with no statement of business associate status anywhere and no scope position taken. One of the two elements the band above asks for is present, which is a grade below it and not a grade at the floor.
Vendor Published

Healthcare Map analyses run on deidentified data, and Komodo says that analytical environment is HIPAA compliant. Its privacy notice covers product users and healthcare professionals' information, and treats deidentified data as outside its personal information rights.

That position does not cover every dataset a customer brings into Marmot. Business associate status, an agreement a customer can read before signing, and subcontractor terms are not described for the wider service. Tokenization and the reidentification ban protect the core data asset, while identifiable customer data has no stated contractual home.

BB on Security Certifications and Trust CenterA recognized certification is named in the vendor own material without the artifact, or with a scope or renewal question the buyer has to raise. A recognized certification here is an audit of how the vendor operates security, such as SOC 2, HITRUST or ISO/IEC 27001. A certification has a scope and a clock, and both are part of this grade. A certification the vendor attaches to its data center, hosting facility or cloud provider, rather than presenting as its own, is the host's and does not count here.
Vendor Published

Komodo's trust center shows a SOC 2 Type II program, with a 2025 report for its HCP product and a 2026 report for Sentinel, both available on authorization, so their boundaries and exceptions are not public. It also lists annual third party penetration testing, disaster recovery, centralized identity management and cyber insurance.

These audits cover Komodo's own operations, not just Amazon's infrastructure. Whether they reach the current Marmot applications is not stated, and the GDPR and CCPA references describe privacy duties rather than security assurance.

CC on FDA and Regulatory StatusNo device claim is made and the product is scoped accordingly. Most administrative and operational products sit here and are not penalized for it, because this axis grades the appropriateness of the positioning rather than possession of a clearance. A stated limit on use, such as research use only or a clause that the product gives no medical advice, beside marketing that sells the product for clinical decisions also sits here, with both quoted, because the buyer has to ask which one governs.
Vendor Published

Marmot supports population analytics, trial feasibility, treatment pattern research and commercial planning, and it is not a diagnostic tool at the point of care. Its outputs can feed research and regulatory submissions, which does not make the platform FDA authorized.

Komodo's status as a CMS Qualified Entity concerns healthcare data and provider performance reporting, a different regulatory role. No device clearance or statutory exclusion is cited for Marmot, and the validity of any study built on it rests on that study's design and data.

BB on AI Governance and Bias DisclosureA governance framework with named process behind it, such as certification to an artificial intelligence management standard, or material written for a customer own review committee to evaluate the product with. A framework covering the vendor whole portfolio counts when its scope reaches this product; what it cannot supply is the product specific evaluation the band above asks for. Published results within the one group where the performance of this product is most at risk also sit here: evaluation on the question this axis asks, short of the comparison across groups the band above requires. A medical device clearance or certification, and the quality system behind it, does not reach this band on its own; it is graded under regulatory status. A regulator reviewed plan governing how the model may change does reach it.
Vendor Published

Saved methods, reusable cohort definitions, query logging, versioned code and an output evaluation gate govern how analyses are built, and teams can see the logic behind an answer and keep the same assumptions for a scheduled rerun. The Healthcare Map methodology also names its coverage limits, including uninsured people and Kaiser Permanente patients in California, and says race and ethnicity data are incomplete.

Those are useful controls and stated data boundaries. Error rates for Marmot itself across demographic groups are not compared, and the approval process for a model change is not laid out.

CC on AI Liability and RecourseMechanisms exist that let someone challenge an output, such as audit trails, source traceability or review before commit, with nothing standing behind the output and no route for the harmed party.
Vendor Published

Marmot keeps source traces and the code behind each answer, so a user can check and correct an output. Komodo's provider correction and appeal route belongs to its Qualified Entity reporting, and nothing says it extends to Marmot analyses.

The master agreement disclaims accuracy warranties and expressly excludes damages from incomplete or inaccurate AI output, caps general liability at 12 months of fees, and offers an intellectual property indemnity that does not cover a wrong clinical or commercial conclusion. The correction tools are concrete; a remedy for a harmful answer is not.

Integration and Deployment
CC on EHR and Interoperability DepthIntegration is claimed through standards or a middleware layer with no system named and nothing to verify.
Vendor Published

MapEnhance adds EHR, laboratory, genomic and social determinants data to the claims based Healthcare Map, and customers can combine their own datasets with Komodo's environment. Data teams reach it through Snowflake, Databricks and Python workflows.

Those are analytics connections rather than writeback to a clinical record, and no production Epic or Oracle Health interface with read and write behavior is named. The core job is longitudinal analysis across sources, not editing a patient's chart, and the data linkage and export serve research rather than two way clinical exchange.

BB on Deployment Model and Data ResidencyOptions and residency are stated with isolation or the processing path left open.
Vendor Published

Marmot runs in a browser and also works from a customer's notebooks, development tools and data platforms. The subprocessor inventory places Amazon hosting, Snowflake storage and the named language model suppliers in the United States, and Marmot commits to isolating each customer's data.

The wider corporate inventory includes international affiliates and support providers, so not every product's processing is confined to one country. A Marmot specific map of cross border paths, backup regions and any private cloud or on premises option is not given, and the territory in an order form governs access rather than where processing happens.

Commercial
CC on Commercial TransparencyNo price is published and the posture is discoverable: a buyer can establish how the product is sold and what drives the cost before contacting the vendor. Most of the index sits here.
Vendor Published

Subscriptions and services are set in order forms rather than a public price table. The portfolio spans the Marmot applications, MapLab Enterprise, MapEnhance data and custom analytics, so the products and data rights chosen define each purchase, and permitted users and territory are written into the agreement.

No subscription rate, seat minimum or implementation schedule is listed. Standard terms call for payment within 30 days and let agreed travel and expenses be billed for services, and a sales demo is the way in.

BB on Setting and Specialty CoverageCoverage is named with validation behind part of it.
Vendor Published

Life sciences teams use Komodo for clinical development, health economics, medical affairs and commercial work. Trial uses include population sizing, feasibility and external comparator cohorts, and other work tracks treatment pathways and utilization. Named customers include Merck, Alnylam, Novo Nordisk, argenx, Dompé and Zevra, alongside the patient group Fight CRC, and the company also serves payers, providers, consultancies, agencies and financial investors.

The claims based population leaves out some uninsured and nonparticipating groups, so broad therapeutic coverage does not mean equal visibility into every patient population, and each cohort depends on its codes and observation windows.

Alternatives

Alternatives to Komodo Health

Other products that do the same kind of work as Komodo Health, and what each one does.

OM1 is a Boston real world data and AI company that Richard Gliklich, a physician, founded in 2015. It sells to drug and diagnostics makers, health systems and payers, and the companies it names include Amgen, AbbVie, GSK, Johnson & Johnson, Hologic and Exact Sciences.

Frekil builds AI agents that run observational real world evidence studies for pharmaceutical, biotech and contract research teams on data they already hold or license, from protocol through final report.

Aetion makes the Aetion Evidence Platform, software for designing and running real-world evidence studies on health data.

ConcertAI's Accelerated Clinical Trials platform helps pharmaceutical sponsors and contract research organizations design, plan and run oncology studies.

Truveta is owned by 30 US health systems that contribute de-identified health records every day to a shared research dataset.

Real world data company built on records patients choose to contribute.

Commercial

Pricing

Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.

Entry Price Pricing Basis BAA Tier Implementation Source
Enterprise quote
Subscription access and licensed data rights are set in an order form. The agreement identifies products, permitted users, territory and service scope. Numeric rates and minimum commitments are not listed. The Healthcare Map uses deidentified patient data. Business associate terms for identifiable data a customer brings are not stated. Professional services and implementation charges depend on the engagement. Agreed travel and expenses may be reimbursable; no standard implementation amount is stated. Vendor Published

The portfolio includes the Marmot applications, MapLab Enterprise, specialty data through MapEnhance and custom analytics. Enterprise agreements define which software and data rights are included. No self serve price table or numeric entry subscription is listed. Standard payment terms are 30 days, and the master agreement's length is separate from each order's subscription.