Healthcare Administrative Automation
H

HiLabs

HiLabs cleans healthcare data that other systems then rely on, and its centre of gravity is the provider directory. Founded in 2014 by chief executive Amit Garg and Neel Butala, based in Bethesda, Maryland, it employs around 340 people across four continents and has raised roughly 41 million dollars, including a 39 million dollar Series B in March 2024 led by Denali Growth Partners and Eight Roads Ventures with F-Prime Capital.

The platform is MCheck, which ingests, cleans and enriches data for health insurers across provider data accuracy, clinical results, payment accuracy and value based care. The provider directory product validates and enriches more than 60 attributes per clinician, from phone numbers and specialties to whether a practice is accepting new patients, by cross checking signals across thousands of sources on a recurring cadence rather than through periodic one off cleanups. The company states the directory solution is live in most United States states and analyses data covering more than 80 percent of the country's healthcare providers, and it works with CAQH, whose provider data it combines with its own models.

The evidence position is unusual for an administrative product. The company's method has been validated in peer reviewed research published in the Journal of the American Medical Association and referenced in a MedPAC report to Congress. The published work found that more than four in five physicians listed in health insurer directories had inconsistent entries, which bears directly on whether the provider directory provisions of the No Surprises Act are being met.

The portfolio has since widened. MCheck NetworkIQ was deployed with a national behavioural health organisation serving millions of members. In August 2026 the company announced MCheck Intelligent Outreach, an agentic voice platform that calls providers to confirm and update their directory details, deployed by one of the ten largest United States health plans. Growth leadership includes president and chief growth officer Amir Desai and chief growth officer Robert Renzi.

AI Health Index verifiedAugust 8, 2026
Compare HiLabs with other vendors
Founded
2014
Headquarters
Bethesda, Maryland
Website
www.hilabs.com
Categories
healthcare-admin-automation, rcm-and-prior-auth, vbc-intelligence
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

The models do the work. Reconciling conflicting statements about a clinician across thousands of sources into one record that is probably correct is an inference problem, not a matching problem, and the company positions explicitly against automation that simply propagates existing errors faster.

Held below A because the output compounds into an asset that is not a model. A validated view of more than 80 percent of United States clinicians becomes the reference the next inference is checked against, and a competitor with equally good models and no accumulated corpus would resolve less confidently. That is the moat is the dataset case, and here it is a direct consequence of the models rather than a substitute for them.

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

Two quite different autonomy positions sit in one company and a buyer should separate them.

The data work is automated at scale: records are validated, corrected and enriched on a recurring cadence without a person reviewing each one, which is the only way the volume is tractable. Nothing published describes a confidence threshold at which a record is routed to a human.

The outreach product announced in August 2026 goes further. An agent telephones a provider's office to confirm details, which means an automated system now initiates the contact that resolves the ambiguity rather than only inferring across sources. The company frames this as beginning from trusted data so calls are targeted confirmations rather than open discovery, which is a sensible constraint. What is not published is what the agent does when the person who answers contradicts the record.

BB on Model and Technology TransparencyThe approach or the suppliers are named without the version and update discipline behind them.
Vendor Published

Operationally specific in the ways that matter to a buyer. The company states the number of attributes validated per provider, more than 60, describes cross checking across thousands of sources, states the refresh cadence and the ability to process ad hoc batches, and quantifies coverage at more than 80 percent of United States clinicians.

The stronger disclosure is methodological: the approach has been described in peer reviewed literature, which means an outside reader can examine how the resolution works rather than taking a platform claim on trust. That is rare in administrative software.

What is absent is accuracy expressed as a rate. No precision or recall figure for the corrections themselves was located, and a claim of processing clinical data 100 times faster than people is a speed claim carrying no accuracy denominator.

CC on Model Supply Chain DisclosureThe architecture is described and no provider is named.
Vendor Published

One distinction materially lowers the risk on the flagship product and should be stated rather than assumed, because a reader scanning this index would otherwise apply the wrong frame. Provider directory data is business information about professionals, covering their names, locations, specialties, contact details and network participation, and it is not patient information, so the core product carries a far smaller privacy exposure than most records here.

A vendor whose primary holding is professional business data should not be graded as though it held charts. What changes the picture is where the platform extends. The company states it handles clinical results and value based care data and is expanding into medical record data, and those are patient information in the ordinary sense, governed differently and carrying the full set of questions this axis asks.

No retention, encryption or model training position was located for either side, and nothing separates the two: a buyer cannot tell whether the controls described for directory work extend to clinical content, or whether the same models and infrastructure serve both. Nothing else is enumerated either, with no hosting arrangement or sub processor list found.

Ask which products touch patient data, whether the infrastructure is shared with the directory product, retention for clinical content specifically, and for a sub processor list covering the clinical side.

AA on Clinical and Operational EvidencePeer reviewed or independently evaluated performance, prospective and multi site where the claim requires it, with the method available to read.
Third Party Estimated

The strongest evidence position of any administrative vendor in this index, and it is earned in an unusual way.

The company's data was not merely validated by outside researchers; it was the instrument in peer reviewed research published in a leading medical journal, which found that more than four in five physicians listed in insurer directories had inconsistent entries. That work was then referenced in a MedPAC report to Congress. So the platform was accurate enough to be trusted as the measuring device in published research on a national policy question, and the finding entered the policy record.

That is the same shape this index credited when a vendor's software was the selection instrument in trials that changed treatment guidelines: contributing the instrument outranks publishing your own accuracy. Deployment corroborates it, across most states, national insurers and a top ten health plan.

One honest qualification. The published work establishes the scale of the problem the company sells into. It does not by itself establish how much of that problem the product fixes, and no before and after directory accuracy measurement was located.

CC on AI Safety and PHI StewardshipGeneral assurances of privacy and security that do not answer the questions artificial intelligence raises: what is retained, what reaches a model, and what happens to it there.
Third Party Estimated

Graded on an honest basis, with a distinction that materially lowers the risk on the core product and should be stated rather than assumed.

Provider directory data is business information about professionals, their names, locations, specialties, phone numbers and network participation. It is not patient information, so the flagship product carries a far smaller privacy exposure than most records in this index.

That changes where the platform extends. The company states it handles clinical results and value based care data and is expanding into medical record data, and those are patient information in the ordinary sense. No retention, encryption or model training position was located for either side.

Regulatory and Compliance
CC on HIPAA and BAA PostureCompliance is claimed without the underlying document, or the published privacy notice covers the website rather than the service that handles patients.
Third Party Estimated

Graded on an honest basis. No compliance statement or agreement posture was located in this pass.

Agreements clearly exist with national insurers and a top ten health plan. The point worth establishing is which product a given agreement covers, since the provider data work and the clinical data work sit under different obligations, and a plan contracting for directory accuracy is in a different position from one sending clinical results.

CC on Security Certifications and Trust CenterControls are described with an outside check behind them, such as independent penetration testing on a stated cadence, but no attestation against a recognised framework.
Third Party Estimated

Recorded honestly and provisionally: the dedicated trust and security search this index requires was not run in this pass, and no attestation was encountered incidentally.

A vendor operating inside national insurers and a top ten health plan has passed substantial security assessment, so the absence reflects what was retrieved rather than what exists.

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 penalised for it, because this axis grades the appropriateness of the positioning rather than possession of a clearance.
Third Party Estimated

No device pathway applies and none is claimed. Cleaning administrative data is not a clinical determination.

The regulatory context is nonetheless the reason this product exists, and it is unusually specific. Federal law requires health plans to verify provider directory information at least every 90 days and to act on updates within two business days, and a subsequent Medicare Advantage statute narrows the focus to directory accuracy with civil penalties for repeated inaccuracies. State regulators can fine plans directly. So the buyer is purchasing compliance with a named and enforceable obligation rather than an efficiency, which is a different commercial position from most of this index and one a buyer can size precisely.

CC on AI Governance and Bias DisclosureResponsible artificial intelligence is committed to in policy language with no evaluation behind it. Most of the index sits here.
Vendor Published

No subgroup analysis, error rate or monitoring policy was located, and the distributional question here is sharper than the abstract framing suggests.

Directory errors are not evenly distributed. They concentrate among small independent practices, behavioural health clinicians and providers outside large integrated systems, because those are the practices without the administrative staff to keep rosters current. Those are also precisely the clinicians patients struggle most to reach, which is why the phenomenon has a name and has drawn congressional attention in mental health specifically.

A resolution system works best where source signals are abundant and consistent, which describes large systems rather than solo practices. If accuracy improves fastest where it was already best, the directory becomes more reliable in exactly the parts of the network patients had least trouble with. Nothing published reports accuracy by practice size or specialty. That single breakdown would be the most informative disclosure the company could make.

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.
Peer Reviewed Publication

The method is described in peer reviewed literature, which means an outside reader can examine how the entity resolution works rather than taking a platform claim on trust, and that is rare in administrative software where the usual standard is a marketing page.

Around it the operational scope is stated with unusual precision: the number of attributes validated per provider, cross checking across thousands of sources, the refresh cadence, the ability to process ad hoc batches, and coverage quantified as a share of the country's clinicians. A buyer can tell what is checked, how often and against how much. Held at C because accuracy is never expressed as a rate.

No precision or recall figure for the corrections themselves was located, and a claim of processing clinical data a hundred times faster than people is a speed claim carrying no accuracy denominator. Precision is the number that matters for this output, because the product's job is to change a record: a wrong correction to a directory entry sends a patient to a clinician who does not practise there, or marks a practising clinician as out of network, and the person who suffers is a patient who never sees the system that caused it.

Recall matters equally, since an uncorrected stale entry persists silently. No warranty, indemnity or remediation commitment was located. Ask for precision and recall on corrections, and what review sits between a proposed correction and a published directory.

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

The relevant systems are payer side, provider data management, credentialing and directory platforms, rather than the electronic health record, and the record is graded on that basis rather than penalised.

The notable integration is with the industry provider data utility, whose collected data the company combines with its own models, which is the right architecture: use the existing attestation channel as one signal among thousands rather than treating it as truth. Stated expansion into medical record data would change this axis materially and has not yet. No interface standard or named platform integration was located.

CC on Deployment Model and Data ResidencyA single hosted option with location implied rather than committed.
Third Party Estimated

Described only as a cloud platform. No hosting region, retention schedule or customer controlled option was located.

The company operates across four continents with a substantial engineering presence outside the United States, which for payer data is a question a buyer will ask directly and which public material does not answer.

Commercial
DD on Commercial TransparencyNothing a buyer can establish before a sales conversation. A published pricing claim contradicted by evidence also grades here.
Third Party Estimated

Nothing published: no price, no mechanism, no unit of sale.

The unit is the interesting question here because the plausible bases differ sharply. Per provider record under management, per attribute validated, per outreach call completed and a flat enterprise licence would each behave very differently for a plan whose network is growing. The outreach product adds a second dimension, since a call that reaches nobody still costs something to attempt.

Against that, the value is unusually easy to size, because the obligation is statutory and the penalties for repeated directory inaccuracy are defined. A buyer can price this against a known exposure rather than against a productivity estimate.

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

Wide on the payer side and specialty agnostic by construction, since a directory contains every clinician a plan contracts with.

Stated reach is substantial: live in most states, analysing data covering more than 80 percent of United States clinicians, with national insurers, a top ten health plan and a national behavioural health organisation among named deployments. Behavioural health is a pointed inclusion given that directory inaccuracy is worst in that specialty.

The product set extends beyond directories into clinical results, payment accuracy and value based care, so the company is broadening from one data domain into several. Held at B because the buyer is almost exclusively the health plan, with providers appearing as the subject of the data rather than as customers, and because coverage is United States only.

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
Not published. Cloud platform sold to health plans, now spanning provider data accuracy, network intelligence and agentic outreach. Not located. Establish which product an agreement covers, since provider directory data and clinical results sit under different obligations. Not published. Deployment involves ingesting the plan's existing provider data and connecting to source signals rather than installing software at a site. Third Party Estimated

Nothing is published: no price, no mechanism, no unit of sale. The unit is the question worth pressing, because the plausible bases behave very differently. Per provider record under management scales with network size. Per attribute validated scales with how thorough the plan wants to be across the 60 plus attributes.

Per completed outreach call introduces a third basis entirely, and a call that reaches nobody still costs something to attempt, so who bears that is a real term rather than an edge case. A flat enterprise licence removes all of it and shifts the risk to the vendor. Ask which applies, and ask separately for the directory product, the network intelligence product and the outreach platform, since they were launched years apart and address different budgets.

The compensating advantage is that value here is unusually easy to size. The obligation is statutory, verification is required at least every 90 days, updates within two business days, and the Medicare Advantage statute attaches civil penalties to repeated inaccuracy. A buyer can price this against a defined regulatory exposure and against what manual verification currently costs them, rather than against a productivity estimate.