Roger Healthcare

Referral summarisation, chart pre fill, ambient capture, the real time prompting about missed items, and completion of the OASIS and routine notes are all model output. No human scribe tier, no coding services layer and no underlying EHR that the AI is a feature of. The company positions itself as a documentation platform sold to agencies, and the RPA sync is plumbing around the model rather than the product itself.

The oversight question here is not whether a human reviews but WHEN, and the ordering is the reason this is C rather than B. Roger's described sequence is that it fills out the entire chart in the agency EMR within minutes and the clinician then reviews the results without changing workflows. That is write first, review second. The pattern this index credits is the opposite, best shown by ModMed, where suggestions populate downstream workflows only once the provider approves.

Two real mitigations belong alongside it. The OASIS is not submitted to CMS until the clinician signs it, so a draft sitting in the EMR is not yet a regulatory filing. And the real time prompting during the visit is a genuine oversight strength that most of the lane lacks entirely, catching omissions while the clinician is still with the patient and can still ask. No confidence threshold, abstention behaviour or escalation path is published.

Graded B for naming its architecture, which most of this lane will not do. Roger describes a multi stage AI pipeline trained on millions of rows of home health visit data, and separately names its EMR sync as a proprietary robotic process automation system. Both are real architectural statements that tell a buyer what they are getting and what to test.

It also gives a concrete worked example of what domain training buys: when a patient says they walked to the mailbox yesterday, the system reads that as implying specific functional scores for ambulation, which is the kind of inference a general model would miss. Held at B because no model card, named model, accuracy figure or evaluation methodology exists, and the headline 80 percent time saving is described as proven without any study, denominator or comparator behind it.

The corpus and the architecture are described and no party is named, which places this mid band. The vendor states a multi stage pipeline trained on millions of rows of home health visit data, and separately names its record system synchronisation as a proprietary robotic process automation layer.

Both are real architectural statements, and the corpus description implicitly answers a question competitors dodge: this model was trained on real assessment data, which is why it outperforms a general scribe on these forms. That makes provenance the question rather than an accusation.

Establish whose data that was, on what basis it was used, and whether the buyer's own agency content joins it, because a specialist model trained on real records is only as defensible as the permissions behind the records. On enumeration there is nothing: no model or model family, no foundation model provider, no hosting arrangement and no sub processor list. One content type needs its own answer.

The product captures photographs of medication bottles to extract prescription data, and a medication label carries the patient's name, frequently their address, the prescriber and the pharmacy, so the image contains identifiers well beyond the drug. Ask for retention and training positions stated separately for label photographs.

One genuinely useful data point and no study behind it. New Day Healthcare, reported at 17,000 patients across six states, has rolled Roger out company wide, which is a named enterprise agency at real scale and is stronger than the anonymous testimonials most of this segment offers. Against that, the 80 percent time saving is asserted as proven with no cohort, baseline or method, and the remaining evidence is unattributed clinician quotes.

On the standing precedent in this index that scale is not evidence, this stays at C. One thing a buyer should check directly: a competitor's own comparison page lists New Day Healthcare as ITS customer while Roger publishes a company wide rollout with the same agency. That may mean parallel evaluation, a switch, or different business units, but both vendors cannot be the sole documentation system there and the discrepancy is worth resolving with the agency rather than with either vendor.

No retention schedule, audio deletion commitment, encryption detail, de identification practice or training use statement was located.

The two concerns raised in the earlier assessment both stand and both are now more concrete.

The training data question first. The model is described as trained on millions of rows of home health visit data. A buyer should establish whose data that was, on what basis it was used, and whether their own agency's content joins it. That is not a hostile question: a specialist model trained on real assessment data is exactly why this product outperforms a general scribe on these forms, and the provenance still has to be answerable.

The access surface second, and it is the sharpest version of this in the category. The integration is a proprietary robotic process automation layer that holds credentials to the agency record system and fills in every field of the chart. Six other vendors in this lane reach the chart through the interface, but they generally deposit a note. This one populates an entire regulated assessment, field by field, under stored credentials. So the vendor holds a durable credential to a system containing every patient that agency serves, not only the ones it documented.

A third category is new and specific. The product captures photographs of medication bottles to extract prescription data. A medication label carries the patient's name, often their address, the prescriber and the pharmacy, so the image contains identifiers well beyond the drug. Nothing published states how long those images are held or whether they are used for improvement.

Ask for retention per content type, the training provenance, and the credential model.

No health privacy statement, business associate agreement reference or compliance page was located in the retrieved materials.

The earlier assessment's reading remains the fair one. Agreements plainly exist, since the platform records patient encounters and writes into record systems at named enterprise agencies, and no agency of that size onboards a vendor handling protected health information without one. The absence is far more likely a website gap than a compliance gap.

What the grade records is that a prospective buyer cannot establish terms before a sales conversation, and that a vendor whose entire product involves recording patients in their homes and holding credentials to an agency's record system has not published a single line about the contractual basis for it. For a company of this size that is a straightforward thing to fix and doing so would move this axis immediately.

One scope point is worth raising when the agreement is produced. The platform now spans more than visit documentation. The intake product processes inbound referral packets, creates patient records automatically in the record system, and runs eligibility checks against a large number of payers. Referral processing means the vendor holds protected health information about people who have not yet been admitted and may never be, and eligibility checking sends identifiers onward to payers. An agreement drafted for ambient documentation may not anticipate either.

Ask for the agreement, its scope across documentation and intake, the position on prospective patients whose referral is declined, and the subprocessor list for the eligibility path.

No named or dated certification, audit report, penetration testing statement or trust centre was located.

The earlier assessment identified why this matters more here than for a typical scribe, and the second pass confirms the architecture that drives it. The integration is a proprietary robotic process automation layer holding credentials to the agency record system, which it drives to fill in every field of the chart.

That access model is what an examination would need to cover, and the questions it raises are not the ones a standard scribe assessment answers. Where are the credentials stored and how are they protected at rest. Whose account does the automation operate under, and does it appear in the record system's audit log as that person. Can an agency distinguish an automated write from a human one after the fact. What happens on staff departure, when a credential should be revoked but an automation depends on it. And what is the blast radius if the vendor's environment is compromised, given the credential reaches the whole record system rather than the visits the vendor documented.

Those are answerable questions and none of the answers is published.

The customer base raises the expectation rather than lowering it. The vendor states enterprise agencies use the platform, and enterprise procurement normally requires a report. The company is small, which is a fair mitigating consideration on timing but not on the access model, since the access model is the same at any headcount.

Ask what external testing has been performed, the credential storage and rotation design, and the audit trail an agency can see for automated writes.

No clearance claimed and none required. The regulatory weight sits with the payment and survey regime rather than device oversight, and the earlier assessment named it precisely: the assessment items including the functional series drive both quality reporting and case mix payment, so a scoring inference drawn from conversation lands in survey findings and payment integrity.

The second pass confirms the product completes exactly those items and adds two things a buyer should weigh.

The first is the vendor's stated position on human review, and it is unusual enough to record plainly. The vendor markets being fully AI driven as a differentiator, contrasting itself with services that use people to quality check notes and framing that review as a delay of a day or more. That is a coherent product argument about speed. It also means the only review between an inferred functional score and a submitted assessment is the clinician's own, in a domain where a direct competitor markets certified coder verification on every note as its principal control. The two vendors in this segment have taken opposite positions on the same question, and a buyer should decide which they want rather than treat either as the norm.

The second is scope. The product generates a structured plan of care from the visit's findings, which is a step past recording an assessment, and the sibling intake product generates diagnosis codes described as maximising reimbursement.

The functional scoring point from a peer applies here too and is worth putting directly: ask whether the system ever revises a clinician's score downward, and in what proportion of cases.

Mixed, with a real positive and a real gap. The positive is that Roger's compliance mechanism runs FORWARD rather than backward: it prompts the clinician about undocumented items while they are still in the home, which prevents an omission rather than querying it later, and reduces the QA back and forth the segment is built around.

Its revenue framing stays on the right side of the coding gradient this index tracks, because the stated mechanism is throughput, serving more patients without adding staff, rather than coding intensity, and on the standing precedent that speed metrics are not coding metrics that does not belong on the gradient. The gap is disclosure.

No fairness statement, subgroup analysis or accent and dialect performance breakdown was located despite multi language capture being sold as a feature, and the workforce in home health is among the most linguistically diverse in US healthcare.

Two passes located no accuracy or error figure, no published limitations and no warranty, indemnity or remediation commitment, with the headline claim of an 80 percent time saving described as proven while no study, denominator or comparator is given. Calling an unpublished result proven is a stronger form of the asserted evidence problem recorded elsewhere in this lane.

The architecture creates a distinct and sharper version of a finding this index has recorded once before, and a buyer should size it deliberately. The integration is a proprietary robotic process automation layer holding stored credentials to the agency record system, and it does not merely deposit a note: it populates an entire regulated assessment instrument field by field. Two consequences follow.

First, the vendor holds a durable credential to a system containing every patient the agency serves, not only the ones its product documented, so the access surface is the whole record rather than the encounters in scope. Second, where software acts under a named person's identity the audit log stops being a reliable account of who entered what, and for a regulated assessment that drives payment and quality reporting, the ability to establish whether a field was entered by a clinician or generated by software is exactly the evidence a later dispute or audit would need. Establish whose credentials are used, whether the automation's entries are separately identifiable in the record audit trail, and what the vendor commits to when a field is wrong.

Functionally deep and architecturally brittle, and the vendor says so plainly, which is the reason for the B rather than a lower grade. The sync is described as a proprietary robotic process automation system operating in near real time, and it works in both directions: pulling attachments, schedules and referral packets out of the EMR before the visit and filling in all fields of the chart after it.

Reading referral documents out and writing the full assessment back is more than almost any RPA implementation in this lane attempts. The costs of that architecture are the standard ones and a buyer should price them in: RPA drives the user interface rather than a sanctioned data channel, so it breaks when the EMR changes its screens, and the audit trail shows an automated session rather than an integration. NO EMR IS NAMED. For a home health agency the systems that matter are WellSky, MatrixCare, HCHB, Axxess, Devero and Netsmart, and which of them are supported today is not published.

No hosting region, residency option, subprocessor detail or offline behaviour was located. Delivery is a cloud service with a mobile capture application and a robotic process automation connector into the agency record system.

Two questions are specific to this product rather than generic.

Offline behaviour is the first, as the earlier assessment noted. Capture happens in patients' homes, frequently in rural areas and inside buildings with poor reception, and the workflow explicitly continues after the visit with additional dictation. Nothing published states what the application does when the signal drops mid visit, what is held on the handset until it syncs, whether it is encrypted at rest there, or what happens if a device is lost with unsynced visits on it. Peers in this segment now treat offline capability as a selling point and describe it directly. This vendor does not mention it.

The second follows from the connector. Because integration works by driving the record system's interface under stored credentials rather than through a permissioned interface, the vendor's environment holds a durable path into the agency's system. So the residency question is not only where audio and transcripts sit. It is also where the credentials sit, and what network path the automation runs over to reach a system that may itself be hosted elsewhere.

A competitor comparison indicates integration with a single major post acute record system, against a broader footprint elsewhere in the segment. Confirm which systems are supported before assuming coverage.

Ask for the hosting region, the subprocessor list, the model provider, the offline and on device retention model, and where credentials are held.

No price, unit of pricing, tier structure or trial is published anywhere, and the only route in is a booked demo.

The one price signal available is a customer testimonial stating that competing products were more expensive and did less. That is a positioning claim from an unnamed source rather than a disclosure, and it is not credited.

Home health only, and deep within it. The product covers the comprehensive assessment rather than a summary of the conversation, and it names the instruments, which is this index's test for real domain depth as opposed to a template library: the OASIS including the M1800 series and GG functional scores, the Plan of Care, wound measurements and medication reconciliation, alongside routine visit notes.

It explicitly targets the failure modes generic scribes hit in this setting, including reading a list of fifteen medications and handling regulatory forms. Multi language capture is supported, with no language count published. Held at B rather than A because coverage is one care setting in one country, with no hospice, no facility based care and no breadth beyond home health.