Purpose-Built AI: from Theory to Practice

by Rowan Report | Feb 12, 2026 | Artificial Intelligence

by Isaac Greszes, Eleos

Purpose-Built AI

From Theory to Practice

This 4-part series has outlined how to evaluate, test, and use AI solutions, emphasizing outcome relevance, workflow fit in regulated environments, architectural scalability, and governance discipline. That framework was intentionally rigorous. In a market crowded with pilots and proofs of concept, it reflects the reality that AI outcomes are not accidental; they are the result of deliberate design choices.

This final chapter shares a real-life story of AI implementation using the Polaris AI Engine.

A Reference Implementation

One example of how these principles are applied in practice is Eleos’ Polaris AI engine.
Polaris was developed over more than five years to support regulated conversational care. Rather than relying solely on general-purpose language models, it combines commercial-grade multimodal infrastructure with proprietary clinical intelligence layers that encode documentation logic, reasoning patterns, and safety heuristics.

Key elements of this approach include:

Layered architecture, separating foundational AI capabilities from clinical logic and governance controls.
Expert-led refinement, with licensed clinicians continuously validating and updating clinical rules.
Application-layer tuning, allowing the system to improve without retraining on customer data.
Governance-by-design, with explicit boundaries around data use, monitoring, and risk management.

Clinical Control

Importantly, Polaris is not positioned as a fully autonomous system. Clinicians remain in control, using AI as a collaborative tool that reduces administrative burden while preserving clinical judgment.

This design reflects a broader principle: in regulated care environments, trust and adoption depend as much on restraint and transparency as on technical capability.

Your content goes here. Edit or remove this text inline or in the module Content settings. You can also style every aspect of this content in the module Design settings and even apply custom CSS to this text in the module Advanced settings.

Applicability in Care at Home

Care at home workflows differ across home health, hospice, and other palliative care settings. Documentation standards, visit structures, and regulatory requirements vary. Validation within each context remains essential.

At the same time, platforms built to handle high-variability conversational care share structural advantages when entering care at home environments; they:

Are designed to operate in unstructured, field-based settings.
Encode clinical reasoning rather than relying on generic text generation.
Incorporate governance and safety controls suited to regulated care.

For executives navigating pilot fatigue, this distinction matters. Platforms designed as infrastructure — rather than experiments — are better positioned to adapt responsibly as care at home AI adoption matures.

Final Thoughts

AI is here and it’s here to stay. Care at home agencies need to look to AI solutions in order to stay competitive. Knowing which solutions to review, what to look for, and how to move beyond the pilot phase begin with finding Purpose-Built Ai. Many thanks to our friends at Eleos for their expertise on this topic. Read the 4-part series.

# # #

About Eleos

At Eleos, we believe the path to better healthcare is paved with provider-focused technology. Our purpose-built AI platform streamlines documentation, simplifies compliance and surfaces deep care insights to drive better client outcomes. Created using real-world care sessions and fine-tuned by our in-house clinical experts, our AI tools are scientifically proven to reduce documentation time by more than 70% and boost client engagement by 2x. With Eleos, providers are free to focus less on administrative tasks and more on what got them into this field in the first place: caring for their clients.

©2026 by The Rowan Report, Peoria, AZ. All rights reserved. This article originally appeared in The Rowan Report. One copy may be printed for personal use: further reproduction by permission only. editor@therowanreport.com

Purpose-Built AI: Evaluation to Execution

by Rowan Report | Jan 30, 2026 | Artificial Intelligence, Partner News

by Isaac Greszes, Eleos

Purpose-Built AI

From Evaluation to Execution

In part one of this 4-part series, we discussed how care at home agencies can realize the full impact of AI software that goes beyond the testing period. The best way to do this is to find purpose-built tech and evaluate AI solutions for real-world outcomes.

In Part two of this series, we outlined how care at home leaders should evaluate AI solutions — emphasizing outcome relevance, workflow fit in regulated environments, architectural scalability, and governance discipline. That framework was intentionally rigorous. In a market crowded with pilots and proofs of concept, it reflects the reality that AI outcomes are not accidental; they are the result of deliberate design choices.

This article examines what execution-ready, purpose-built clinical AI actually looks like in practice — and why certain platforms are structurally better positioned to deliver sustained value in care at home settings.

Market Tenure is a Weak Signal

As AI adoption accelerates across healthcare, many organizations default to a familiar proxy for confidence: market tenure. Vendors with early pilots, a growing logo list, or proximity to large EHR ecosystems are often assumed to be safer bets.

In emerging AI categories, however, tenure can be misleading. Early adoption frequently reflects experimentation rather than readiness. Platforms may perform well in narrow pilots while masking deeper limitations in clinical depth, scalability, or governance that only surface during broader rollout.

Design is a Better Measure

For care at home leaders under pressure to move beyond pilots, the more reliable question is not how long a vendor has been in the market, but how the system was designed to operate under real-world clinical and regulatory constraints.

Purpose-Built AI

What it Means Under the Hood

Generic AI tools often struggle in care at home environments. Here, it is worth examining what distinguishes purpose-built clinical AI at a structural level.

Purpose Built AI Evaluation to Execution

Clinical-grade platforms share several characteristics:

Clinical reasoning embedded in the system, not inferred from prompts. The AI reflects how clinicians assess, prioritize, and document care — rather than simply summarizing conversations.
Structured outputs aligned to documentation and reimbursement requirements, ensuring that generated content is usable without extensive manual correction.
Safety-aware interpretation of sensitive language, particularly in areas related to risk, decline, or end-of-life care.
Governance mechanisms baked into the architecture, including transparency, monitoring, and clearly defined limits on data use.

Conversational Care

Why are conversational care settings more challenging? Clinical insight derived from spoken interactions rather than structured inputs present some of the most complex challenges for AI systems.

Conversational care requires the AI to:

Interpret unstructured dialogue occurring in non-clinical environments
Distinguish clinically meaningful information from casual conversation
Recognize implicit risk signals and contextual nuance
Translate narrative interaction into structured, compliant documentation

Added Challenge

Behavioral health and substance use disorder care represent some of the most demanding examples of this complexity. Systems that perform reliably in these environments must handle variability, sensitivity, and regulatory scrutiny simultaneously.

This matters for care at home leaders because many of the same challenges — environmental variability, role-based documentation requirements, and safety-sensitive language — are present across home health and hospice workflows.

Next Steps

As organizations move from evaluation to execution, several questions can help distinguish platforms capable of delivering sustained value:

Can the vendor clearly explain how clinical reasoning is encoded in the system?
Are outputs structured to align with documentation, compliance, and reimbursement needs?
How is safety monitored and governed over time?
What mechanisms exist to adapt workflows without destabilizing operations?
Where does ROI typically emerge once AI is embedded into daily practice?

Answering these questions does not guarantee outcomes – but it significantly reduces the risk of prolonged pilots with limited impact.

Final Thoughts

The next phase of AI adoption in care at home will favor platforms built for durability, governance, and clinical trust. For leaders, the challenge is no longer whether AI can help, but how to select systems designed to deliver value beyond the initial pilot phase.

Understanding how AI was built — not just what it promises — is now a prerequisite for confident execution. Come back next week for the fourth and final installment in this serious where we will discuss a real-world implementation example.

# # #

About Eleos

Purpose-Built AI: Architecture, Scalability, Security

by Rowan Report | Jan 23, 2026 | Artificial Intelligence

by Isaac Greszes, Eleos

Purpose-Built AI for Care at Home

Architecture, Scalability, and Security

Part two focuses on AI architecture, scalability and security.

Architecture and Scalability Across the Tech Ecosystem

AI does not operate in isolation. It sits within a broader ecosystem of EHRs, compliance programs, quality initiatives, and IT infrastructure.
For care-at-home organizations, long-term outcomes depend on whether an AI platform can:

Adapt to evolving documentation and regulatory requirements
Scale reliably during census fluctuations
Integrate cleanly with existing systems
Improve over time without creating operational drag

Health informatics research increasingly highlights risks such as model drift — where AI performance degrades as populations, workflows, or clinical practices change — reinforcing the need for continuous monitoring rather than one-time deployment.

purpose-built AI architecture and scalability

Vendors with limited clinical depth or brittle configurations may show early promise in pilots, but often struggle to sustain efficiency and ROI at scale.

Security, Governance

and the Link to Long-Term Value

HIPAA compliance remains foundational, but AI introduces additional governance considerations related to transparency, accountability, fairness, and ongoing risk management.

Healthcare organizations increasingly evaluate AI vendors based on:

Independent security and privacy assessments
Clear contractual boundaries around data use
Explicit retention and deletion policies
Documented processes for monitoring AI behavior over time

Expectations

Recent federal regulation, including the ONC’s HTI-1 Final Rule, formalizes new transparency and risk-management expectations for AI-enabled clinical systems — extending well beyond traditional privacy frameworks.

Emerging standards such as ISO 42001, focused on AI management systems, reflect a broader shift toward formal governance of AI in high-risk domains like healthcare. While adoption is still evolving, these frameworks provide executives with a useful lens for assessing vendor maturity.

Strong governance is not only a risk-mitigation strategy — it is a prerequisite for sustaining outcomes, protecting organizational reputation, and maintaining provider trust.

A Practical Takeaway

AI has demonstrated the potential to reduce administrative burden, improve documentation quality, and deliver measurable ROI in healthcare — including regulated, care at home settings.

However, results are not guaranteed. They depend on evidence-backed design, workflow alignment, scalability, and governance discipline.

For care-at-home leaders, the most reliable path to value is not adopting AI quickly, but evaluating it rigorously — with a focus on how the technology is built, validated, and governed.

For organizations navigating pilot fatigue, the critical shift is not testing more tools, but selecting platforms designed for scale, governance, and long-term operational impact.

# # #

This is part 2 of a 4-part series. Read part 1 and come back next week for part 3, “From Evaluation to Execution.”

About Eleos

Purpose-Built AI for Care at Home

by Rowan Report | Jan 15, 2026 | Artificial Intelligence, Home Health, Hospice, Outcomes

by Isaac Greszes, Eleos

Purpose-Built AI for Care at Home

How Care at Home leaders can move beyond AI pilots

Care at Home is increasingly turning to AI to address documentation burden, clinician burnout, and regulatory pressure. While AI has the potential to address these issues and more, practical results remain uneven, leaving agencies with a lot of experimentation, but little clarity on actual value.

Evaluating AI solutions should focus on real-world outcomes, how the solution fits into your existing workflow, whether the software is scalable, and how it handles changing regulations. You should also look for AI solutions that are built for care at home (purpose-built). This series of articles will help you make informed, risk-aware decisions about AI adoption.

AI is Coming Fast

Home health and hospice leaders are navigating a difficult balance: persistent workforce shortages, rising provider burnout, expanding documentation requirements, and increasing regulatory scrutiny — all within thin operating margins.

At the same time, AI has moved quickly from experimental to strategic. Many organizations are now evaluating AI not just for productivity, but for operational and administrative efficiency, clinician experience, compliance readiness, and financial performance.

And the stakes are high

Early results across the market have been inconsistent. Some organizations report meaningful reductions in administrative burden and a clear return on investment. However, others struggle to find value after adoption. The difference often lies not in whether AI was adopted, but how it was designed, supported, and governed.

The pilot problem

As AI adoption accelerates, many organizations find themselves caught in extended pilot cycles — testing multiple tools without committing to the operational changes required for scale. While pilots can validate technical feasibility, they rarely provide the consistency or measurement discipline needed to demonstrate sustained ROI in regulated care at home environments.

Quality over Quantity

Why the right evidence matters

In today’s AI market, product demonstrations are easy to produce. Documented outcomes are not.

Executive leaders should expect vendors to demonstrate real-world impact, supported by customer data, third-party validation, or peer-reviewed research. Credible AI partners should be able to explain how their results translate to care at home — and where limitations exist. The challenge is not the lack of information from pilots, but the lack of evidence those pilots results can be reproduced, measured, and sustained, in a care at home setting.

Objective Evidence that Matters

When evaluating AI platforms, leaders should look for evidence related to:

Documentation efficiency, such as reduced time per visit or faster note completion
Operational ROI, including quicker billing readiness or reduced rework
Compliance support, such as documentation completeness or audit preparedness
Provider experience, including reduced perceived administrative burden
Care outcomes, including patient engagement and satisfaction

AI solutions can impact efficiency and burnout. But, these outcomes are highly dependent on whether the solution was built for care at home, the quality of implementation, how easily it will integrate into your workflow, and governance. If a vendor cannot explain how results were achieved and whether they are reliable and repeatable outside the pilot, the vendor and the solution should be evaluated carefully.

General Purpose AI

And inconsistent results

Many AI tools marketed to healthcare organizations rely on general-purpose language models designed for tasks like summarization, chat, or content generation — not for producing structured clinical notes aligned to regulatory and reimbursement requirements.

Home health and hospice documentation often includes:

Clinical observations made in non-clinical environments
Structured requirements tied to reimbursement and regulation
Risk-sensitive language related to safety, decline, or end-of-life care
Significant variation across disciplines, visit types, and patient contexts

Where generic AI breaks down

In these settings, AI tools based on general-purpose language models introduce risks related to accuracy, hallucinations, bias, privacy, and workflow fit — because they were not designed to operate within structured clinical, regulatory, and reimbursement frameworks.

In practice, organizations report that the additional oversight required to validate or correct AI-generated output can reduce — or even negate — anticipated efficiency gains, limiting adoption and ROI. As a result, organizations often remain stuck in pilot mode — investing time and effort in validation without achieving the scale or consistency required for meaningful return.

The right question

When evaluating an AI solution, the right question is not whether the AI tool can record a conversation and translate it into notes or whether the tool can reduce documentation, but whether it can consistently support high-quality clinical documentation at scale without increading burden or creating compliance risks.

Purpose-Built AI

What it means and why it drives operational impact

In care at home environments, purpose-built AI should be evaluated less as a point solution and more as foundational infrastructure — one designed to support regulated clinical workflows consistently over time.

Many AI platforms label themselves as “purpose-built,” but leaders must look past marketing language to truly scrutinize the way the technology is designed and deployed. In regulated clinical environments, purpose-built AI typically incorporates:

Domain-specific clinical intelligence, informed by real documentation patterns
Provider involvement in defining structure, logic, and validation criteria
Structured outputs aligned to required note components, in addition to free-text summaries
Grounding mechanisms that reduce fabricated or misattributed content
Privacy-conscious data handling, with explicit limits on data retention and reuse

Research consistently shows that providers prefer AI systems that function as collaborative tools — preserving human oversight while reducing administrative load — rather than fully automated systems that completely bypass clinical judgment. These characteristics directly affect whether AI improves documentation time, supports compliance workflows, and earns provider trust — all prerequisites for driving ROI.
These design choices are what allow AI systems to move beyond experimentation and begin delivering durable efficiency, compliance support, and clinician adoption at scale.

# # #

This article is part 1 in a 4-part series. Come back next week for “Scalability, Security, and Governance.”

About Eleos

Eleos Navigates Eligibility Risk

by Rowan Report | Aug 21, 2025 | Admin, Artificial Intelligence, Clinical, Home Health, Hospice, Medicaid, Partner News

Eleos Navigates Eligibility Risk

FOR IMMEDIATE RELEASE

Contact: Amanda Wells

awells@sloanepr.com

Eleos Launches AI Scanner to Navigate Medicaid Eligibility Risk in Real Time

The new OBBBA AI scanner uses Eleos’ ambient AI technology to alert providers of patient eligibility changes, preserving revenue and ensuring care continuity amid sweeping Medicaid policy changes

BOSTON, MA, Aug. 20, 2025 — Eleos, the leading AI platform in post-acute care, today announced the launch of the OBBBA (One Big Beautiful Bill Act) AI scanner, the first real-time tool to proactively detect potential changes to Medicaid eligibility during client sessions. The OBBBA AI scanner uses Eleos’ purpose-built ambient AI scribing technology to inform providers about changes that may impact coverage, giving them time to act before Medicaid coverage lapses. The tool was launched in response to sweeping Medicaid funding cuts and eligibility rule changes.

Eligibility Check

Providers can select Medicaid-related “themes” to track such as housing status, diagnosis updates, or life events like marriage or aging out of eligibility. The OBBBA scanner captures contextual clues that could trigger changes in coverage. Providers use this information to take action to prevent eligibility loss, reduce care disruption and maintain treatment continuity. For care organizations, this means fewer denials and greater revenue stability, as well as better client support.

The OBBBA AI scanner arrives at a critical moment: new Medicaid rules introduce shorter retroactive coverage windows, semi-annual (versus annual) redeterminations and narrowed eligibility criteria — all of which lead to a higher risk of churn, especially for vulnerable groups such as people with serious mental illness and those experiencing housing instability.

“We’re hearing from leaders across the country that Medicaid redetermination changes are already causing confusion and fear among clients and providers alike. The OBBBA AI scanner gives providers the earliest possible warning via real-time insights so they can protect coverage and avoid treatment disruptions, ensuring clients continue to receive necessary and life-saving care. This kind of provider-first technology is at the core of Eleos.”

Alon Joffe

Co-founder and CEO, Eleos

Embedded seamlessly within the Eleos Documentation experience, the tracker works in tandem with providers’ existing workflows, requiring no additional software or manual data entry.

Industry leader sees Eleos scanner as critical tool

“OBBBA has created significant uncertainty for the behavioral health sector, and organizations need every possible advantage to navigate it. Properly deployed, purpose-built AI tools help organizations navigate an ever-changing landscape while also promoting the health and well-being of clients and communities.”

Chuck Ingoglia

President and CEO, National Council for Mental Wellbeing

Rationale

The OBBBA AI scanner builds on Eleos’ mission to free care providers from administrative burdens and enable better, more data-informed care. Deployed in over 200 organizations in 30-plus states, Eleos is the most-used AI solution in behavioral health, substance use disorder (SUD) treatment and post-acute care. Its suite of AI-powered documentation and compliance solutions has been proven to reduce documentation time by more than 70%, double client engagement and drive 3-4x better treatment outcomes.

For more information about the OBBBA AI scanner or to request a demo, visit www.eleos.health.

# # #

About Eleos

Eleos is the leading AI platform for behavioral health, substance use disorder, home health and hospice. At Eleos, we believe the path to better care is paved with provider-focused technology. Our purpose-built AI platform streamlines documentation, simplifies revenue cycle management and surfaces deep care insights to drive better client outcomes. Created using the industry’s largest database of real-world sessions and fine-tuned by our in-house clinical experts, our AI tools are scientifically proven to reduce documentation time by more than 70%, boost client engagement by 2x and improve symptom reduction by 3-4x. With Eleos, post-acute care providers are free to focus less on administrative tasks and more on what got them into this field in the first place: caring for their clients.