Andy AI

Capture, chart drafting, the OASIS itself, wound and medication documentation, the QA pass over the full history and physical, and coding output are all model generated. The human contribution is described as review and confirmation rather than production: clinical experts review, clinicians confirm codes in the app. One qualification a buyer should test.

The company describes the product as human reviewed by clinical experts and the coding as coder backed, without stating how many charts those reviewers touch or how much they change. If reviewers are substantively rewriting output rather than verifying it, this is a services business with an AI front end and the grade should move, which is the same question this index puts to every vendor carrying a human layer.

A review gate is described at two distinct points, which is better than most of this lane manages. Documentation is drafted by AI, confirmed by the clinician, then synced to the EMR, and coding output requires clinician confirmation inside the app before it is billing ready. A separate QA layer described as human review by clinical experts sits between draft and chart.

The companion app is explicit that it does NOT generate an OASIS on the device and that the agency must connect its EHR for the completed chart to appear, so nothing reaches the record through an unsanctioned path. Held at B rather than A because the reviewers are not credentialed in the way Lime Health names a credentialed home health coder verifying every chart by default, no threshold or abstention behaviour is published, and it is not stated whether QA review is applied to every chart or sampled.

No model card, named model, accuracy methodology or evaluation design located. The performance claims are stated as outcomes rather than as measurements: 45 percent more detailed charts, 24 percent more patient data captured, 45 percent more conditions identified, double the clinician productivity. Each of those needs a denominator and a comparator to mean anything, and none is given. A third party review site describes the accuracy as superhuman, which is that site's language rather than the vendor's, and is not evidence.

Nothing identifies any party in the chain: no model or model family, no foundation model provider and no sub processor list was located in two passes, with the only statement being that captured data uploads to a cloud described as meeting health privacy requirements, which names a property rather than a provider. The architecture makes the omission concrete rather than abstract.

The vendor states the assessment is not generated on the device, and that an agency must connect its record system to the vendor's cloud to receive completed output, so capture, upload and generation are separate stages and each involves a transfer to parties nobody has named. The content type raises the stakes further. This application captures images as well as audio, which in home based care means wound photographs taken inside a patient's home.

A photograph is identifiable in ways a transcript is not, it may capture the patient's body, their living conditions and other people present, and it persists as an image rather than as text that could be redacted. Nothing published states how long photographs are held, whether they are stored separately, or whether they are excluded from any model improvement. Ask for a sub processor list, and for retention and training positions stated separately for images and for audio.

No study, controlled evaluation, third party assessment or named customer agency located, and no deployment count of any kind. Every figure is vendor reported and uncited: one hour saved per start of care chart, 90 minutes reduced to a 15 minute review, export in under five days, a 7 percent case mix increase, double productivity. The funding and founder credentials are verifiable through Y Combinator, Crunchbase and Dealroom and are real, but investor validation is not clinical evidence. The company also declined to submit products for a CB Insights analyst briefing, which is neutral in itself but means no third party analyst view exists either.

The only commitment located is that captured data uploads to a cloud described as meeting health privacy requirements. No retention schedule, no audio deletion commitment, no de identification practice and no statement on whether customer content is used to train or improve models was found.

Two things make that gap larger here than for a clinic based scribe.

The application captures images as well as audio. In home health those are wound photographs, and they are taken inside a patient's home. A photograph of a wound is identifiable in ways a transcript is not, it may capture the patient's body, their living conditions and other people present, and it persists as an image rather than as text that could be redacted. Nothing published states how long photographs are held, whether they are stored separately from the note, or whether they are excluded from any model improvement.

Second, the mobile application's store disclosures indicate that contact information and user content are collected and linked to user identity. Those labels are vendor declared summaries rather than a policy, and they describe the app rather than the platform, so they should not be read as a finding in themselves. They are a reason to ask the vendor directly what is linked to whom and for what purpose, rather than to infer from a store listing.

The processing architecture makes the questions sharper. The vendor states the application does not generate the assessment on the device, and that an agency must connect its record system to the vendor's cloud to receive completed output. So capture, upload and generation are separated, and each stage has its own retention question.

Ask for the retention schedule for audio and images separately, and for the training position in contract language.

HIPAA compliance is stated consistently and specifically, with the cloud described as HIPAA compliant in the product materials and in both app store listings. No business associate agreement terms, availability or pricing were published. This lands on the common rung: compliance claimed, BAA unpublished.

The agency rather than the individual clinician is the contracting party here, which is a structurally healthier position than the free and self serve tools in this lane where a clinician can start recording before any agreement exists.

A second pass located no named or dated certification, no audit report, no penetration testing statement and no trust centre. Security is described through the phrase health privacy compliant cloud, which is a regulatory claim rather than an attestation, and there is no certifying body for that rule in any case.

The exposure profile is what makes the absence worth pressing rather than accepting as a young company norm. This is a mobile first product used outside any controlled environment. Clinicians carry it into patients' homes, capture audio and photographs there, and upload from whatever connection is available. The devices are personal or agency issued phones rather than managed workstations inside a facility, and the physical setting has none of the controls a clinic has.

An independent examination is the mechanism that tells a buyer how that surface is governed: how devices are enrolled and revoked, what is retained on the handset before upload and for how long, what happens when a phone is lost with unsynced visits on it, how access is provisioned for agency staff, and whether any of it is logged in a way the agency can audit.

None of that is answerable from published material, and an agency remains the covered entity carrying the obligation.

Fairness is due to the stage of the company, which is recent and venture backed, and a full attestation programme is a material cost at that size. This grade reflects what a counterparty can verify before contracting rather than a judgement that controls are absent. Ask for whatever external testing has been performed, and specifically for the device and upload security model.

No clearance claimed and no device pathway attaches. This axis does not read as an absence, because a real regulatory framework governs the product and the earlier assessment named it correctly: OASIS drives both the quality reporting that determines agency star ratings and the case mix weighting that determines episode payment, so failures land in survey findings and payment integrity rather than in device safety.

The second pass adds a finding that is subtler than the usual coding observation and worth setting out carefully, because the vendor is not doing anything improper.

The vendor states that its system is trained to score OASIS on what is safe for the patient rather than on what the patient currently does. That is the correct convention. The functional items are scored on safe ability, and clinicians commonly score habitual activity instead, which is a well known source of inaccuracy. Applying the correct rule is a genuine quality improvement.

It also runs one way. Scoring on safe ability rather than observed activity systematically produces lower functional scores, lower functional scores raise the case mix weighting, and the vendor separately reports agencies seeing a seven percent case mix increase and the system finding forty five percent more conditions. So the interpretation the product applies is simultaneously the more accurate one and the more remunerative one, and nothing published distinguishes how much of the reported increase is corrected under scoring and how much is drift.

That is the question to put, and it is answerable: ask whether the system ever revises a clinician's scoring downward, and in what proportion of cases.

Credited alongside: coding is described as requiring clinician confirmation in the app, which is a stated review gate rather than an implied one.

The reason for the C is one number: agencies using Andy are reported to see a 7 percent case mix increase. Under PDGM, case mix weighting determines what a 30 day home health episode pays, so this is the home health instance of the coding intensity gradient this index tracks, and it sits at a larger unit than most. The pattern is now visible at three scales: E and M level determines what a visit pays, DRG determines what an admission pays, and case mix determines what an episode pays.

Real counterweights belong in the same paragraph and are why this is C rather than D. Andy frames coding as accurate and defensible, requires clinician confirmation before anything is billing ready, and the underlying claim is that more complete capture of conditions genuinely present produces a more accurate case mix, which under a system that pays for documented clinical complexity is a legitimate argument rather than only a commercial one. There is no contingent pricing and no human removed.

Separately, no fairness statement, subgroup analysis or accent and dialect disclosure was located, which matters in a segment where the workforce and the patient population are both linguistically diverse and the recording happens in the patient's home.

Two passes located no accuracy or error figure, no published limitations and no warranty, indemnity or remediation commitment. The published claims are stated as outcomes rather than measurements, including more detailed charts, more patient data captured, more conditions identified and doubled clinician productivity, each expressed as a percentage with no denominator, comparator or method.

One of those deserves separate attention because it is presented as a benefit and is equally readable as a risk signal. More conditions identified may mean the system captures what a rushed clinician would have missed, which is the intended reading, or it may mean the system documents conditions the clinician did not assess.

In home based care that distinction is not academic: documented conditions drive payment grouping, and a system that reliably increases the number of conditions on a chart is doing something that regulators examine closely. Nothing published lets a buyer tell which is happening, because no accuracy or confirmation rate is given for the conditions the system adds.

The question to ask is how many of the additional conditions a clinician confirms on review and how many are removed, since that ratio is the difference between improved capture and inflated documentation. Ask for that ratio, and for what the vendor commits to when a condition is added in error.

The architecture is genuine write back rather than transfer. The completed chart including the OASIS is generated into the agency EHR, the agency connects its EHR to Andy's cloud as a deployment step, and the companion app deliberately does not produce an OASIS on the device precisely so the record of truth stays in the EHR. That is the right shape.

The gap is that no EHR is named anywhere. Lime Health, the closest comparator in this segment, names its six post acute systems explicitly. Andy says only that it is compatible with leading home health EHRs and that specific systems require confirmation during onboarding. For an agency running WellSky, MatrixCare, HCHB, Axxess, Devero or Netsmart, whether their system is supported today is the first question, and the answer is not published.

No hosting region, residency option or subprocessor list was located, and nothing establishes whether a third party model service processes the encounter or what it retains.

The architecture is now clearer and it is split in a way that matters. The vendor states plainly that the application does not generate the assessment on the device. Capture happens on the phone, data uploads to the vendor's cloud, generation happens there, and completed output reaches the clinician only once their agency has connected its record system to that cloud. So processing is server side by design and the mobile application is a capture and upload client.

That makes offline behaviour a first order question rather than a detail, and it is the specific gap on this record. Home health visits happen in patients' homes, frequently in rural areas and inside buildings with poor reception. Nothing published states what the application does when the signal drops mid visit: whether capture continues locally, what is held on the handset until a connection returns, how long it stays there, whether it is encrypted at rest on the device, and what happens if the visit is never synced.

That is both an operational question and a security one, and peers in this lane answer it. One states its application functions offline and syncs later and is deployable in facilities without site wide wireless coverage, and treats that as a selling point. Another is built for field work in the same setting. This vendor says nothing.

Ask for the hosting region, the subprocessor list, the model provider, and specifically the offline behaviour and on device retention model.

No price published. Quotes are customised by agency size, clinician volume and EHR integration needs, and access runs through a demo. What lifts this above the unrated floor is one genuinely useful commercial disclosure: the company states the bundle costs less than agencies typically pay for QA and coding alone, which anchors the buyer to a cost line they already carry rather than to a per clinician software rate.

That is a benchmarkable claim even without a number, and it also tells a buyer what is being displaced. Ask whether the price is per clinician, per chart, per episode or per agency, because in home health those produce very different totals at the same headline rate.

Narrow by design and deep within it. Home health only, spanning start of care through discharge, covering both nursing and therapy disciplines with therapy evaluations and exercise documentation captured during treatment. The depth marker this index looks for is present: it names the instruments rather than counting specialties, working to the OASIS and PDGM and organising the product around start of care as the pivotal chart.

Wound documentation via photograph and medication list reconciliation are the two field specific capabilities. Held at B rather than A because coverage is a single care setting in a single country with no language breadth described, and because it addresses home health only rather than the wider post acute picture.