Cedar Gate Technologies

Predictive modelling is claimed across the analytics layer, and the assets that make this company hard to replace are the payment engines and the data foundation underneath them.

The claims are present and consistently made. Company material describes applying machine learning predictive models to surface cost saving and quality improvement opportunities for sharing with provider partners, and defining cohorts by risk profile using advanced models to target interventions and personalise care. Risk cohorting over combined claims and clinical data is a real inference task and there is no reason to doubt it is performed.

What holds the grade is what happens if the models are removed. A single data lake that ingests, cleanses, normalises and member stitches disparate sources still has substantial value. A bundles engine that automatically converts fee for service claims into a single bundled claim and settles payment between payer and provider still works, because adjudication is deterministic by design and must be. A capitation engine administering agreements against changing eligibility rolls still works. Those are the components that took acquisitions and years to assemble.

One qualification separates this record from others graded at the same level in this lane. Peers name the model driven product: a prescriptive engine, a record review capability, a conversational agent. Here the capability is described at the level of marketing language without a named product behind it, so a reader cannot identify which shipped feature the models sit inside. Graded C on consistency with the lane rather than because the evidence matches the stronger records in it.

Ask which named product contains a predictive model, what it predicts, and what it achieves.

This platform executes financial transactions automatically, which makes it the most consequentially autonomous record in this lane, and it is autonomous in a direction the axis is not usually pointed at.

The analytics and care layers behave conventionally. Insights surface to plan and provider staff, care management supplies workflow to people who then act, and nothing touches a patient without a person.

The payment layer is different in kind. The bundles engine automatically converts fee for service claims into a single bundled claim and facilitates and manages payment between payers and providers, and the capitation engine administers agreements against changing eligibility. That is software adjudicating and settling money without a human approving each transaction, at a volume the company describes in the millions of bundles. It is also the correct design, because adjudication at that volume cannot be manual and determinism is precisely what makes it trustworthy.

The oversight questions are therefore financial rather than clinical, and none is answered in material located. What happens when an episode is adjudicated incorrectly, who detects it, whether reconciliation is automated or manual, what the dispute path is for a provider who believes a bundle was settled wrongly, and whether adjudication logic changes are versioned and communicated before they take effect are all unaddressed.

An error here does not harm a patient. It underpays or overpays a provider, potentially systematically and potentially for months before anyone notices, and the party least able to detect it is usually the one being underpaid.

Graded C.

Ask what reconciliation and dispute process sits behind adjudication, and how logic changes are versioned and notified.

Architecture is described with reasonable clarity and the inference layer is described only in adjectives.

The architectural account is usable. A single data lake ingests, cleanses, normalises and member stitches disparate sources, a composable application layer sits above it covering analytics, care and payment, and the payment components are named and their function stated precisely, including automatic conversion of fee for service claims into a single bundled claim. A technical evaluator can follow how data moves and what each component does.

The models are not described at all. Company material refers to applying artificial intelligence algorithms and machine learning predictive models, and to defining cohorts using advanced models, without naming a single one, stating what it predicts, or publishing any accuracy, calibration or validation figure. A dedicated pass located no model card, no retraining cadence and no drift monitoring account.

The gap between the two halves is itself informative. When a vendor can describe its adjudication engine to the level of what it does to an individual claim but describes its predictive capability as advanced models, the disparity usually reflects where the engineering documentation actually exists.

The benchmarking layer has a related omission. Comparison against a 12 million life database rests on how that database is constructed, weighted and matched to a comparable peer group, and none of that methodology is published, which makes a benchmark result difficult to interpret and impossible to challenge.

Graded D.

Ask what the predictive models are, what they achieve, and how benchmark peer groups are constructed.

The chain is closed at every link, including the infrastructure link that most vendors in this lane at least name.

A dedicated pass located no sub processor register, no named cloud or infrastructure provider, no third party component inventory, no data licensor and no subcontractor list. Data is described as warehoused in the cloud without identifying whose.

The acquisition history is the substantial question. Four businesses were absorbed between 2018 and 2020, and one of them brought a bundled payment engine with a lineage the company describes as spanning more than forty years. Software of that age carries accumulated dependencies, and whether component inventories across the four acquired estates have been consolidated, or collected at all, is not described. For an engine that adjudicates payment, the provenance of the logic and the components it runs on is not an academic question.

The benchmark database is a data supply chain in its own right. Twelve million member lives came from somewhere, and neither the contributing sources nor the terms under which their data may be used in a commercial comparison product are identified.

The labour supply chain is disclosed in outline and not in substance. A team in Nepal is named in company material, which is more candour than most, and nothing describes what that team does, what data it can reach or under what controls.

Ownership adds a governance dimension. Under a global parent since late 2025, decisions about which components are consolidated or retired sit with a new owner, and nothing describes how such changes reach customers.

Graded D. Ask for the sub processor register, the infrastructure provider, and what the offshore team accesses.

A long list of named customer results, one analyst recognition, and headline figures large enough to require the method that is not supplied.

The better evidence is the specific and modest kind. A cardiovascular care network reports a 26 percent reduction in spending per cardiac patient after moving to a bundled arrangement built on the adjudication software. A medical university with more than 100 locations reports a 3 percent improvement in comorbid chronic kidney disease patients avoiding hospitalisation, with average savings of 36,100 dollars for those who did. Both are plausible in magnitude, both are attached to identifiable organisations, and the second states its denominator population. A research and consulting firm named the company an innovation leader in the United States population health management market in 2025.

The headline figures are a different matter. A bundled payment case study presented at an industry summit in 2026, alongside two named partner organisations, reports 97 percent patient satisfaction, 91 percent avoided inpatient hospital care, a 60 percent reduction in total cost of care and a 50 percent reduction in surgical rates. Those numbers are extraordinary. A 50 percent reduction in surgical rates and a 60 percent reduction in total cost of care would each be a landmark result in health services research, and presented together without a denominator, a time period, a comparison group or a stated method they cannot be assessed at all. The most likely explanation is a narrow selected cohort in a single programme, which is a legitimate result described in a way that invites a much larger reading.

No peer reviewed publication and no external validation of the data or measure logic was located.

Ask for the denominator, period and comparison group behind each headline figure.

The data engineering is described concretely, and the benchmark database raises a secondary use question the published material does not close.

The custodial description is better than most in this lane. Data is ingested from disparate sources, cleansed, normalised into standardised files and stitched directly to members, with automated enrichment producing usable information. Member stitching is the hard part of aggregation and naming it as a distinct step indicates the problem is treated seriously rather than assumed away.

The question is the benchmark asset. Buyers are offered comparison against an anonymised database described as covering 12 million member lives, which is presented as a benefit and is a real one, since a cost and utilisation figure means little without a peer group. What is not described is where those 12 million lives came from, and the plain reading is that they came from customers. If so, the questions are what contractual permission allows one customer's data to inform another customer's benchmark, what anonymisation standard was applied and by what method, whether the process was externally reviewed, and whether a customer may decline to contribute while still consuming.

The word anonymised is doing considerable work in that description and is not defined anywhere located. Claims and utilisation data at member level is among the more re identifiable data that exists, and the difference between a statistical aggregate and a de identified record set is the difference between a low risk and a meaningful one.

Model data governance is unaddressed in the ordinary way, with no statement on whether customer data trains models or whether learning crosses customer boundaries.

Graded C. Ask what permits contribution to the benchmark database, what anonymisation standard applies, and whether a customer can decline.

The posture is asserted and nothing located substantiates it.

Company material states that a single integrated platform means one team minimising security and privacy risk, which is a marketing proposition rather than a compliance disclosure. A dedicated pass located no certification, no audited report, no template business associate agreement, no breach notification window, no liability cap position, no audit rights statement and no data return provision.

Business associate status is the obvious posture for an organisation holding identified claims, clinical, eligibility and benefits data for health plans, providers and employers. The absence is of evidence rather than of the obligation.

Two features of this business make the gap heavier than it would be elsewhere. The payment engines adjudicate claims and move money between payers and providers, which brings financial data handling alongside clinical data handling and typically attracts a separate set of obligations that nothing published addresses. And the workforce is described as spanning offices in three United States cities, a large remote population and a team in Nepal, so protected health information is plausibly accessible across jurisdictions. Nothing located describes how offshore personnel access is governed, screened or restricted, and that is a question a covered entity is required to be able to answer about its business associates.

Graded D on absence of disclosure rather than on any adverse finding.

Ask for the template agreement, the breach notification window, and how personnel outside the United States access protected health information.

A dedicated pass located no security certification, no audited report and no trust surface of any kind.

Nothing was found covering the healthcare assurance framework, service organisation control reporting at either assurance level, information security management certification, payment card standards, penetration testing cadence or vulnerability disclosure. No trust centre, security page or compliance portal was located. The nearest thing to a security statement in company material is a marketing proposition that using one integrated platform from one vendor means one team minimising security and privacy risk, which is an argument for consolidation rather than evidence of a control environment.

This is recorded as an absence of published evidence rather than a finding that no certification exists. A company of roughly 1,000 employees adjudicating claims for health plans and processing data for large provider organisations would in the ordinary course hold audited reports, because its customers' procurement processes would require them. The finding is that a prospective buyer cannot obtain any of it without entering a sales conversation.

The contrast within this lane is stark and worth stating. Health Catalyst publishes which certification covers which named product with framework versions and coverage periods. Lightbeam publishes a certification and its renewal history. Persivia publishes a certification and its tier. Here there is nothing to compare, and the payment adjudication component means this vendor handles financial transaction flows that most of those peers do not.

Graded D.

Ask which certifications and audited reports exist, their scope across the module set, and whether any are available before a confidentiality agreement.

Outside device regulation, and regulated in the financial and programme direction instead.

Analytics, benchmarking, care workflow and payment adjudication are administrative and financial functions rather than diagnosis or treatment, so no clearance is required and none is claimed.

The regulatory weight sits elsewhere and is genuine. Software that adjudicates claims and settles payment between payers and providers operates inside a body of state insurance regulation, prompt payment requirements and contractual arrangements with defined timelines, and a capitation engine administering eligibility touches enrolment rules. Support for a mandated federal episode payment model introduced in March 2026 places part of the product inside a programme where participation is compulsory for affected hospitals and where the rules are set by a regulator rather than negotiated.

What is not described is how the company demonstrates conformance in any of those directions. No statement was located covering how adjudication logic is validated against programme rules, how rule changes are absorbed when a federal model updates its specifications, or what assurance a customer receives that a settlement complies with the arrangement it was meant to implement.

That is a narrower question than device clearance and a more immediate one for a buyer, because a mandated programme has deadlines and a vendor that lags a specification change puts its customer out of compliance rather than merely out of date.

Graded C.

Ask how adjudication logic is validated against programme specifications and how quickly rule changes are absorbed.

Two distinct bias mechanisms operate in this product and nothing published addresses either.

The first is the familiar one. Cohorts defined by risk profile using models over claims and utilisation history, then used to target interventions and personalise care, sit squarely inside the documented failure mode where historical spending used as a proxy for health need understates illness in populations who have historically received less care. A patient sorted into a low priority cohort is never contacted and never learns they were sorted.

The second belongs to bundled and episode based payment and is specific to this vendor. Episode pricing depends on risk adjustment, and if the adjustment under compensates for social complexity, comorbidity burden or language need, then providers serving harder populations lose money on bundles that providers serving easier populations profit from. The rational response is to avoid the harder patients, and the mechanism that produces it is arithmetic rather than intent. Analytics that identify which episodes are profitable are the same analytics that identify which patients to avoid, and nothing published describes whether the platform detects or discourages that use.

Benchmarking adds a third consideration. Comparison against a 12 million life database tells an organisation how it performs against a peer group, and if that peer group is not adjusted for population differences, an organisation serving a deprived population will read as an underperformer when it is doing harder work.

A dedicated pass located no bias testing, no subgroup performance figures, no fairness review, no model documentation and no statement of target variables.

Ask what the cohorting models optimise, how episode risk adjustment handles social complexity, and how benchmarks control for population differences.

Recourse is undefined at every point, and this platform carries a direct financial exposure that most records in this lane do not.

A dedicated pass located no indemnification position, no warranty covering model or analytic output, no accuracy guarantee, no service credit regime and no described dispute process.

The distinctive exposure is adjudication. When software converts fee for service claims into a bundled claim and settles payment between a payer and a provider, an error is not a misleading report, it is money moved incorrectly. Systematic error in bundle construction or episode attribution could underpay providers for months across every episode processed, and the aggregate becomes material quickly at the volumes described. Capitation administration against changing eligibility carries the same shape, since paying on stale enrolment produces overpayment or underpayment that persists until someone reconciles it.

The party bearing that risk is not obvious and nothing published resolves it. A provider underpaid through a vendor's adjudication has a contractual relationship with the payer, not with the vendor, so the dispute runs through a party that did not perform the calculation. Whether the payer can recover from the vendor, and on what terms, is precisely the question a buyer should have answered before signature.

The analytic exposures sit alongside it in the usual way: a cohorting model that mistargets intervention resources, or a benchmark that misrepresents performance and drives a bad strategic decision.

Graded D.

Ask what liability attaches to adjudication error, what the reconciliation and recovery process is, and how disputes involving three parties are resolved.

Aggregation across the payer and provider divide is the stated design goal and the architecture matches the claim.

The problem is framed correctly in company material: payers hold claims, providers hold clinical records, and value based care requires both sides working from one view. The platform is built around that, ingesting claims, clinical, eligibility and benefits data into a single lake, then cleansing, normalising, enriching and stitching records to members so both parties see the same performance picture. Integrating virtually any source into a unified foundation is the stated capability, and member stitching across claims and clinical sources is the genuinely hard part of it.

The consumption side is open, with self service configuration of dashboards and reports, tiered access, and automated report distribution, so data that arrives can be shared across an organisation and with provider partners rather than remaining inside one team's screens.

The payment layer adds an integration requirement most platforms in this lane do not carry, since adjudication must exchange data with existing claims systems and partner ecosystems, and robust interfacing is described as a design property.

What holds it below the top is the absence of specifics and of external corroboration. No count of natively supported electronic health record systems, no named integration standard, no description of how identity is resolved technically, no statement on whether integration is bidirectional, and no third party validation of the data of the kind two other vendors in this lane hold.

Ask how many source systems are supported natively, how identity is resolved, and whether findings write back into clinical systems.

Cloud hosting is asserted and nothing else about the deployment is described.

The single located statement is that data is warehoused securely in the cloud. No infrastructure provider is named, no region is described, no tenancy model is stated, no residency position is published, and no recovery objective, availability commitment or failover description was found.

Two characteristics of this business make the omissions matter more than the usual amount. The payment engines adjudicate claims and settle money on schedules that participants depend on, so availability is not a convenience question. A analytics platform unavailable for a day is an inconvenience, whereas an adjudication platform unavailable during a settlement cycle delays payment to providers who have already delivered care, and no commitment covering that was located.

The second is the workforce description. Company material describes offices in three United States cities, a large remote workforce and a team in Nepal. That says nothing definitive about where data is stored, but it does raise the question of where data is accessed from, and the published material addresses neither.

The estate assembled through four acquisitions raises the usual unanswered question about whether separately built applications share one deployment or several, which bears directly on the composable proposition, since a customer adding modules over time needs to know whether it is extending one system or operating two.

Graded D on absence of disclosure.

Ask where data is hosted and accessed from, whether tenancy is dedicated, and what availability commitment attaches to adjudication.

Cost is absent from every published surface. A dedicated pass located no pricing page, no unit of charge, no range, no tiering, no implementation fee position and no minimum commitment.

The composable architecture is the specific reason a buyer needs guidance here and does not get it. The platform is marketed as modular, so an organisation can take data management alone, add analytics, add care management, add one or both payment engines, and build toward an end to end deployment over time. That is a genuine commercial advantage and it makes the cost curve the central question. Nothing published indicates whether modules price independently, whether the data foundation carries a fixed cost that later modules ride on, or whether taking the full suite is cheaper or dearer than assembling it.

The payment layer raises a second unit question that most records in this lane do not have. Adjudication is transactional work, so bundles and capitation processing plausibly charge per bundle, per claim, per member per month or per agreement administered, and an engine described as having processed millions of distinct bundles clearly has an established transactional rate. None is published.

Services compound it again. The company describes managed services and consulting alongside self service technology, and the March 2026 bundled payment launch offers both as explicit options, which is a pricing choice presented without either price.

Ask how modules price incrementally, the transactional unit for adjudication, and the difference between the managed and self service options.

Coverage of payment models is the distinctive breadth here, and it is more unusual than coverage of clinical conditions.

Most platforms in this lane serve one or two arrangements well, typically shared savings and accountable care participation. This one addresses primary care attribution, shared savings, prospective and retrospective bundles, and capitation, across commercial, Medicare and Medicaid lines, with support for a mandated federal episode model launched in March 2026. Bundles and capitation are meaningfully harder than shared savings because both require adjudicating payment rather than reporting on it, and few vendors attempt both.

Buyer coverage extends past the usual pair. Health plans, provider organisations, self funded employers, third party administrators and adviser partners are all addressed, and the employer and administrator segment is served with its own benefits analytics product. Self funded employers are a genuinely different customer with different data, different questions and no clinical operation of their own, and serving them is not a variation on serving a health system.

Clinical coverage follows from the payment models rather than from condition specific content, with bundles built around procedural episodes such as cardiac and musculoskeletal care and analytics addressing chronic disease populations.

What holds it at B is depth evidence. Breadth assembled through four acquisitions raises an unanswered question about how completely the parts function as one product, and no installed base figure by segment was located.

Ask how deeply the acquired applications integrate, and what the customer base looks like by buyer type.