ArcHouse
ArcHouse, made by HospiceMatch, Inc., is software for the work between a referral and an admission at hospice, palliative care, home health and house call providers. It pulls referrals arriving by fax, email, phone, portals and EHR channels into one inbox that deduplicates them, assigns teammates and tracks response times. ArcHouse Intelligence reads faxes, clinical notes, discharge summaries and referral forms, extracts demographics, diagnoses, medications and insurance, checks Medicare and other coverage in real time, and ranks the next actions on each referral, from requesting a history and physical to sending consent forms to a family. In the product view, eligibility checks run and record requests go out while a drafted consent waits for staff approval.
ArcHouse Records connects to health information exchanges through Metriport the moment a referral is created, pulling hospital, primary care, lab and pharmacy records into one chart, and summarizes hundreds of pages into diagnoses, comorbidities, hospitalizations and medications in seconds, with diagnosis specific analysis pitched as support for Medicare LCD criteria and prognostication. Hospice of Davidson County raised referral conversion from 73% to 78% across about 320 referrals in each four month period around its July 2024 rollout, and Hospice of Hope reports 18.8% more patient days year over year.
The security page cites a SOC 2 Type II audit, 256 bit encryption at rest, single sign on and an immutable audit log, with the report and a business associate agreement available to prospects under a nondisclosure agreement. Packaging covers small businesses, multi office groups and enterprises, and pricing comes through a demo.
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
ArcHouse Intelligence does the reading and much of the follow up: it parses faxed referrals, extracts patient, diagnosis, medication and insurance details, summarizes retrieved charts into a structured clinical picture and ranks the next best action on each referral. Those model driven steps sit inside a workflow that also includes a shared inbox, assignments, response time tracking, eligibility checks, electronic signatures and conversion reporting.
An intake team could still move referrals through the queue without the summaries, but the automatic reading of mixed faxes and records, the speed the company sells, would be gone. The homepage claim of document processing ten times faster rests on that extraction.
The AI works alongside the intake team, drafting messages and consents, requesting documents and verifying coverage, while staff review, approve and decide, and every AI action is visible and attributable. In the product view, a drafted consent waits for approval while a Medicare eligibility check runs and a record request has already gone to a hospital, and staff can run a recommended action with one click. Clinicians review the summary before an admission decision. Which actions may proceed without approval, what level of confidence stops an extraction, and how a request sent in error is recalled are not set out.
Three kinds of model work are described: document understanding that reads faxes, clinical notes, discharge summaries and referral forms and extracts fields; clinical summarization that turns hundreds of pages into diagnoses, comorbidities, hospitalizations and medications in seconds; and recommendations that rank next actions by urgency. That separates reading, summarizing and acting in a way an intake leader can follow.
The optical recognition and language models behind them, their training data, versions and validation against hospice charts are not identified, and how the diagnosis specific analysis maps records to coverage criteria is not explained.
Metriport is the named partner for record retrieval from health information exchanges, and the privacy policy describes a Google calendar connection whose data reaches only workspace members and subprocessors. The suppliers of document recognition, summary generation and task recommendations are not named, and no subprocessor list sets out who receives original faxes, retrieved charts or generated text during processing, or on what retention and training terms. The privacy policy's general clause on service providers covers website personal data rather than patient records, so the model path for protected health information stays out of view.
Hospice of Davidson County went live in July 2024 and compared March through June with August through November, leaving out the rollout month: referral to admission conversion rose from 73% to 78% across 324 and 322 referrals, with staff reporting time saved. Hospice of Hope used the records, inbox and request tools and reports an 18.8% rise in patient days year over year.
The homepage adds claims of a 50% shorter path from referral to admission, document processing ten times faster and more than two hours saved per referral, without a source. Neither case separates the AI's effect from faster record access or measures summary accuracy.
Referral packets carry demographics, insurance details and clinical histories pulled from hospitals, primary care, labs and pharmacies. The security page describes 256 bit encryption at rest and TLS 1.2 or later in transit with keys rotated in a dedicated key management service, roles scoped to intake, clinical, billing and leadership work, and referrals locked to a site or region. Those controls govern access.
How long faxes, retrieved records and generated summaries are kept, whether any customer data trains the extraction or summary models, and what happens when a summary omits or misstates a key chart detail are not described.
The security page describes the platform as built to handle protected health information with the safeguards required by the HIPAA Security and Privacy Rules and lists it as HIPAA compliant, with business associate agreements available to prospects under a nondisclosure agreement. The only privacy policy covers the ArcHouse website, contact details, usage data, cookies and a Google calendar connection, and says nothing about referral charts or retrieved records.
The agreement, its tiers and its terms for subcontractors cannot be read before signing. The status is stated, while the document that would show how patient data is protected stays behind the nondisclosure step.
A SOC 2 Type II audit covers the availability, confidentiality and integrity of customer data, and the report is available to prospects under a nondisclosure agreement, with the security team offered for reviewer questions. Controls listed beside it include 256 bit encryption at rest, TLS 1.2 or later in transit, keys rotated in a dedicated key management service, SAML single sign on, multifactor authentication, SCIM provisioning, round the clock monitoring, patching on a fixed cadence and an immutable, exportable audit log. The audit period, the auditor and whether model processing falls inside the boundary are visible only in the gated report.
The products are sold for intake operations and chart preparation, and the Records page goes further, offering diagnosis specific analysis that supports core elements of Medicare LCD criteria and prognostication for hospice eligibility. The medical disclaimer in the terms says HospiceMatch is a software company, not a healthcare provider, that it does not render medical or clinical advice, and that any decision on the appropriateness of treatment is the user's alone. Both stand side by side, a general limit on advice and a marketed prognostication aid, with no device claim and no named regulatory exclusion for the eligibility analysis.
Every AI action is shown with its state and who or what performed it, and clinicians review the structured chart summary before deciding on admission. Those controls make the workflow visible. No governance framework covers how the extraction and summarization models are developed, approved for release or tested, and nothing shows whether summaries miss details more often for some diagnoses, for patients with long or fragmented records, or for referrals from particular channels.
Hospice eligibility turns on diagnosis specific evidence, so how the action ranking and the analysis behind LCD support perform across conditions such as COPD, heart failure and others is the open question.
Intake and clinical staff review the output, and an immutable activity log records every status change, note, message, signature and export with the person or system responsible, filterable and exportable for compliance review. The terms of use, which cover the ArcHouse software as well as the website, disclaim any warranty of accuracy or completeness, make the customer solely responsible for validating all output and reports, and cap liability at the fees paid in the prior 12 months, with disputes going to AAA arbitration under New York law. A wrong summary or a misrouted referral leaves the agency with the log and its own correction.
Metriport, a named data partner, runs the live connection that pulls medical records from health information exchanges into ArcHouse when a referral is created, gathering hospital discharge summaries, primary care notes, lab results and pharmacy medication lists into one chart. Intake also takes EHR referrals into the shared queue alongside fax, email and portal traffic. That is a working read path through named middleware.
No record system is named for writing admissions or summaries back, no hospice EHR such as WellSky or Homecare Homebase appears as a connection, and no interface specification describes the outbound side.
ArcHouse is a hosted application, and the terms say the service is controlled and operated from offices in the United States. Inside the application, permissions follow teams and locations, so a multisite operator can lock referrals, notes and reports to a region or site, and access can follow the customer's identity provider through SAML single sign on and SCIM.
No hosting provider, production region, backup location or customer hosted option is named, and where faxes and retrieved charts are processed by the extraction and summary models is not stated, so location rests on the company's US operation rather than a commitment.
Sales runs through a demo, with packaging described for small businesses, multi office groups and enterprises and solution pages for hospice, palliative care, home health and house calls. The offer bundles unified intake, records retrieval, referral analytics, eligibility checks, consent tracking and the AI actions. No price appears per organization, location, user or referral, and no minimum, module price or charge for interfaces and onboarding is given. A buyer can see the scope of the purchase and the footprint it suits, while the cost depends entirely on the sales conversation.
Care at home is the whole market: solution pages cover hospice, palliative care, home health and house calls, sized for small businesses, multi office groups and enterprises, and the company describes customers ranging from regional hospices to multistate home health networks. The detailed results come from hospice, at Hospice of Davidson County and at Hospice of Hope, which also runs palliative care. Home health agencies and house call programs are named as buyers without comparable case results, so validation is concrete for hospice intake and thinner for the other settings the product targets.
Alternatives to ArcHouse
Other products that do the same kind of work as ArcHouse, and what each one does.
Carefeed is an operations platform for senior living, skilled nursing and long term care, built to connect admissions, staff, families and billing around the resident record and HR system a community already runs.
Valerie Health runs an AI front office for independent physician groups, combining software agents with its own operations staff.
Courier Health gives biopharma teams a shared system for managing patient access to specialty therapies. It connects information from hubs, specialty pharmacies, payers and internal teams so staff can see enrollment progress, authorization barriers, therapy starts and ongoing adherence.
Luminai automates administrative work in large health systems end to end.
Basata connects referral, fax and call work with the systems provider teams use to register patients and schedule visits.
Claim Health sells an AI platform for home-based care agencies (home health, hospice, home care and other post-acute providers) that now presents itself as a care continuity platform running from referral to discharge.
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 |
|---|---|---|---|---|
|
Contact sales
|
The hosted platform is sold by demo and agreement, with packaging for small businesses, multi office groups and enterprises. | A business associate agreement is available to prospects under a nondisclosure agreement, and its tiers and text are not shown before signing. | Onboarding and configuration are described, with no implementation or interface amount stated. | Vendor Published |
No rate appears per organization, location, user or referral, and module, minimum and onboarding charges come through contracting.