Personalized care agent · Chronic pain & behavioral health

The personalized care agent for chronic pain
& behavioral health.

Source-grounded monthly care plans for the QHP. Continuous between-visit signal from the patient. The reasoning architecture was extracted from 100,000+ minutes of integrated chronic pain & behavioral health care delivered in 2025 — five repeatable clinical patterns, encoded as the agent's reasoning surface. Designed for G3002, the CMS ACCESS Model, and the False Claims Act enforcement environment of 2026.

Schedule a documentation & personalization review See the 5 patterns

Operating today at Polaris Therapy, Yakima WA

100K+ min

Integrated pain + BH care delivered in 2025

5 patterns

Encoded as reasoning architecture

G3002 + ACCESS

Reimbursement anchors built in

CDS exempt

21 U.S.C. § 360j(o)(1)(E)

What we are not

Not another AI scribe. Not a wellness app. Not a percentage-of-claims play.

The healthcare-AI landscape of 2026 is crowded. Most of it is the wrong shape for chronic pain and behavioral health. We chose a different surface, on purpose.

An ambient documentation product. Abridge, Epic AI Charting, OpenEvidence, Doximity GPT own encounter capture and medical-evidence search. We sit on top of your EHR via FHIR R4 and operate at the level of monthly care planning — not visit transcripts.
Autonomous. Every output requires synchronous QHP attestation. Every documented fact traces to a source artifact. Every clinical judgment traces to clinician-authored reasoning. Unsupported outputs fail closed.
A wellness app. The patient companion is clinician-enrolled, not consumer-acquired. It captures structured PEG, PHQ-9, GAD-7, sleep, function, adherence, side effects, and flares — and it does not tell patients to change medications.
A percentage-of-claims tool. Flat pricing only: base + per-QHP seat + panel tier. We do not take a percentage of captured reimbursement and we do not pay for referrals. This is upstream of Anti-Kickback Statute analysis, not a safe-harbor argument.
A black box. Source-grounded outputs across the agent's surface. Element-level QHP review at the pre-bill gate. Audit packet ready on demand.
A regulated medical device. We meet all four factors of the 21st Century Cures Act CDS exemption (21 U.S.C. § 360j(o)(1)(E)). We ingest wearable summaries and patient-reported activity/sleep, not raw physiological signals — a deliberate scope decision.

The empirical foundation

The architecture started in clinical observation, not in a lab.

In 2025, our clinical co-founder Dustin — Johns Hopkins-trained counselor (M.S. 2017), owner-operator of Polaris Therapy LLC in Yakima, WA — delivered 100,000+ minutes of integrated chronic pain and behavioral health care across his patient panel. The five repeatable clinical patterns he observed are now the agent's reasoning architecture. Each pattern maps directly to a product component.

01

A substantial fraction of chronic pain patients are over-medicated, not under-medicated.

ObservedPatients arriving on four to six pain-related medications. The actual gap is not another agent but better matching of existing agents to phenotype, consolidation, adherence support, and behavioral integration.

Encoded asThe agent's core decision principle of "what should the clinician try next?" — which often means consolidating, swapping, or adding behavioral interventions, not piling on.
02

Behavioral health is the rate-limiting factor on pain outcomes — not a comorbidity to manage alongside.

ObservedPHQ-9 ≥10 or persistent insomnia produces systematically worse pain trajectories than peers with similar baseline pain severity but stable mood and sleep — even on identical pharmacotherapy regimens.

Encoded asThe integrated chronic pain + BH thesis. One circuit problem, not two — consistent with central-sensitization neurobiology (descending PAG/RVM serotonin/norepinephrine pathways and NMDA-mediated wind-up). What the literature describes mechanistically, the architecture treats as the primary integration target.
03

Information loss across the care team is the dominant clinical risk.

ObservedThe PCP doesn't know what the pain specialist tried. The pain specialist doesn't know what the counselor sees in weekly contact. Mood deterioration on gabapentinoids gets misattributed to depression and stacked with another antidepressant. Opioid-mediated REM disruption worsens pain perception. SSRI-driven sexual dysfunction gets attributed to "stress." Critical signals go unaddressed because no one is integrating the longitudinal multi-modal signal.

Encoded asThe Care Graph and the longitudinal patient companion. Memory across the team, across visits, across months — so no one has to reconstruct the chart from scratch each time.
04

Trajectory is predictable by Week 4.

ObservedWith weekly PROMs and clinical observation, an experienced clinician can predict which patients will improve on a regimen and which will deteriorate roughly 4–8 weeks before the trajectory becomes visible to the prescribing team.

Encoded asOutcome-attainment forecasting at Week 4. Not an AI breakthrough — the encoding of what an experienced clinician already does, made consistent and scalable across thousands of patient-months. The AI moat compounds with the dataset.
05

Documentation completeness predicts clinical continuity.

ObservedEven excellent care is fragile if the reasoning is not documented in a way that supports continuity. New clinicians picking up the chart cannot reconstruct what was tried, what worked, what failed, and why decisions were made.

Encoded asThe compliance & evidence layer — source-grounded facts, element-level QHP attestation, audit-ready packets. Continuity by architecture, not by memory. The same architecture that defends the chart against an audit is what makes the next clinician's job possible.

The architecture is the encoding of what an experienced clinician already does — made consistent, scalable, and documentable. The agent generalizes the pattern recognition; it does not require Dustin in every loop. Pattern recognition is the starting architecture; multi-practice longitudinal data is what converts it to evidence. — XPain.AI thesis

What we ship

Three components. One Care Graph.

A clinician-facing workbench, a clinician-enrolled patient companion, and a compliance-and-evidence layer. Each maps directly to one or more of the 5 patterns above.

Layer 01 · Clinician · Patterns 01, 02, 04

The chronic pain care agent

Synthesizes EHR history, medications, validated PROMs (PEG, BPI, PROMIS, PHQ-9, GAD-7, WHODAS), wearable summaries, and diagnostics into a draft monthly care plan covering pharmacotherapy, behavioral interventions (CBT-CP, ACT, behavioral activation, sleep restoration), monitoring frequency, side-effect watchlists, and stepped-care escalation.

CYP2D6, serotonin-burden, geriatric-risk, MME safety checks
Phenotype-aware combination therapy (Gilron Lancet 2009; pregabalin + duloxetine; LDN)
Outcome-attainment forecasting at Week 4 — encoded clinical pattern recognition, not an AI breakthrough

Layer 02 · Patient · Pattern 03

The patient companion

Clinician-enrolled. Captures structured PEG, PHQ-9, GAD-7, sleep, function, adherence, side effects, and flares — weekly. Surfaces personalized insights from the patient's own data: pain trajectory, medication response, sleep–pain correlation, mood–function relationship.

Validated PROM cadence built for G3002 and ACCESS reporting
RPM/RTM workflows where paired with FDA-cleared third-party devices
Targets >70% weekly PROM completion versus typical 30–40% baseline

Layer 03 · Practice · Pattern 05

The compliance & evidence layer

Enforces evidence rules, blocks incomplete records at the pre-bill gate, and produces audit-ready packets on demand. Coding logic is configured per practice across the relevant workflows and validated by certified coding partners as payer rules evolve.

G3002 / G3003 chronic pain management
APCM, BHI, CoCM where applicable (G0568, G0569, G0570)
ACCESS outcome reporting and Track-2 readiness

How it works

The QHP's monthly chronic care loop, on rails.

A single workflow, repeated every month per patient. The agent drafts; the clinician decides; the patient reports; the next month's plan reflects the response.

Phenotype + historyEHR, medications, PROMs, wearable summaries, diagnostics — assembled and timelined.
Draft planPharmacotherapy, behavioral interventions, monitoring frequency, side-effect watchlists, escalation criteria — all source-linked.
QHP reviewElement-level edit and synchronous attestation. Unsupported claims fail closed.
Patient companionWeekly PEG, PHQ-9, GAD-7, sleep, function, adherence, side-effect signal between visits.
Next month's planOutcome data feeds the next draft. Outcome-attainment forecasting flags Week-4 trajectories that need intervention.

The agent is a synthesis layer, not a decision-maker. Pain medicine requires bedside judgment that current foundation models cannot replicate. The clinician-supervision design is a scientific necessity, not just a regulatory hedge.

Compliance posture

Built inside the regulatory boundary, not outside it.

FY2025 False Claims Act recoveries reached $6.8B+ overall, $5.7B+ in healthcare. We architect for that environment — not for the one before it.

FDA / device classification

21st Century Cures Act · 21 U.S.C. § 360j(o)(1)(E)

We meet all four factors of the FDA CDS exemption: we do not interpret raw signals, we display medical information, we support clinician recommendations, and we enable independent review of the basis for every recommendation. No FDA submission required at current scope.

Privacy & security

HIPAA Privacy & Security Rules · 42 CFR Part 2

BAAs with all practice partners. AES-256 at rest, TLS 1.3 in transit. Audit logs, role-based access, designated Privacy and Security Officers. Heightened protections for substance-use-disorder records.

Anti-Kickback

42 U.S.C. § 1320a-7b(b) · Flat pricing structure

Flat software fees only — no percentage of captured reimbursement, no payment for referrals, no shared claims revenue. Our arrangement does not implicate the Anti-Kickback Statute in the first place. Polaris Therapy MSA at fair market value, reviewed by healthcare counsel.

FCA defensibility

31 U.S.C. § 3729 · Pre-claim review architecture

Source-grounded fact attribution, element-level QHP attestation, synchronous review at the pre-bill gate, audit packet ready on demand. "Human in the loop" is not enough if the human is rubber-stamping. Pre-claim review at the compliance layer is the structural answer.

AI risk management

NIST AI RMF · Foundation-model evaluation discipline

Evaluation, traceability, transparency, accountability, monitoring, and risk controls across the agent's lifecycle. Healthcare AI in 2026 requires both the regulatory posture and the AI safety posture. We architect for both.

What would change scope

Deliberate red lines in product design

If we added autonomous clinical decision-making, direct-to-patient diagnostic claims, raw physiological signal interpretation, or our own hardware, FDA scope would expand. Each is a stated red line. We integrate with FDA-cleared third-party RPM/RTM devices but do not build them.

Compliance receipts

As of May 4, 2026 · Updated continuously

Standard	Status	Note
HIPAA — BAAs in place at deployed sites	Live	Active at Polaris Therapy. Templated for all incoming practices.
CDS exemption posture (21 U.S.C. § 360j(o)(1)(E))	Live	Four-factor analysis on file. Architecture maintains exemption by design.
AKS structural review	Live	Flat pricing across all engagements. Polaris MSA reviewed by healthcare counsel at FMV.
42 CFR Part 2 segmentation	Live	Consent flows and SUD data segmentation in place.
NIST AI RMF baseline	Live	Evaluation, traceability, monitoring across the agent lifecycle.
SOC 2 Type II	In flight	Initiated Q3 2026 (post-batch). 6–9 month attestation cycle.
Cyber, E&O, D&O insurance	In flight	Scaling with deployment count.
HITRUST CSF	Planned	Targeted post-Series A as enterprise health-system contracts emerge.
ONC HTI certification	Planned	Not required for independent-specialty-practice GTM. Targeted post-Series A.

A workbench output

One draft care plan, in the agent's voice.

A representative case typical of an integrated chronic pain + BH practice. Synthetic patient. Real evidence base. The QHP reviews, edits, and attests before anything is documented or billed. Each recommendation is tagged with the pattern from the empirical foundation it applies.

workbench / monthly-care-plan / PT-00713 Draft · awaiting QHP attestation v0.4 · synthetic

patient: 58F · diabetic peripheral neuropathy · 4 yrs

PROMs: PEG 7.3 · PHQ-9 15 · GAD-7 11 · ISI 18 · WHODAS 32

history: prior gabapentin escalation → DC for sedation @ 1800mg. SSRI naïve. No opioid exposure.

phenotype: central-sensitization-mediated, with affective + sleep components

safety: CYP2D6 normal (chart-confirmed) · QTc 412ms · no serotonergic burden

Recommended plan

P01P02Initiate nortriptyline 10mg qHS · titrate to 25–50mg over 4 weeks

Addresses descending NE tone and sleep architecture in a single agent. Combination therapy outperforms either monotherapy at maximum tolerated dose. Single-agent consolidation per Pattern 01; integrated pain-BH circuit logic per Pattern 02.

Source · Gilron et al., Lancet 2009 — RCT, n=56, DPN/PHN

P01Hold further gabapentin escalation

Sedation-limited at 1800mg in this patient. Phenotype favors NE/sleep mechanism over additional Ca²⁺-α2δ binding. Avoid medication stacking absent mechanistic rationale.

Source · prior chart, encounter 2025-11-08

P02Refer to CBT-CP · 6-session protocol

Behavioral component for sensitization-mediated phenotype with affective overlap. Concurrent with pharmacologic titration. Pain + BH treated as one circuit problem, not two parallel tracks.

Source · ACCESS Track 1 behavioral integration guidance

Monitoring · companion-driven

cadence: weekly PEG, PHQ-9, ISI · daily flare/sleep · adherence + side-effect prompts

stepped-care trigger: P04Week-4 PEG reduction <30% → re-evaluate; consider duloxetine cross-titration or low-dose naltrexone augmentation

forecast: attainment-probability output suppressed pending model calibration on Polaris cohort (n<50)

Documentation · pre-bill check

G3002: all elements present · 24 min QHP time recorded · care plan updated · PROMs documented

audit packet: ready · P05 11 source artifacts linked · attestation pending

Every recommendation is source-linked and tagged to one or more of the 5 empirical patterns. The QHP edits or rejects any element. Nothing reaches the chart, the patient, or a claim until attested.

Synthetic example for illustration. Not a real patient. Not a treatment recommendation for any individual. The agent never produces treatment recommendations to patients; recommendations are surfaced to the supervising QHP and reviewed before any clinical action.

The competitive landscape

Where we sit in the 2026 healthcare-AI map.

We are structurally adjacent to the well-funded incumbents, not directly opposed. Different surface, different data, different buyer, different time horizon.

Layer	What they own	What we own
Generalist clinical AI Abridge · Epic AI Charting · OpenEvidence · Glass · Pathway	Theirs: Encounter documentation. Medical-evidence search. Enterprise health-system distribution.	Ours: Monthly chronic care planning, not encounter capture. Independent specialty practices, not enterprise systems.
Voice-first patient agents Hippocratic AI	Theirs: Voice-first patient outreach at scale. Interaction logs. Pharma engagement breadth.	Ours: Longitudinal phenotype-response data inside an integrated chronic pain + BH workflow. Response trajectories, not interaction logs.
Pain-specialty EHRs ModMed · ChartLogic · PrognoCIS · Athena · eClinicalWorks	Theirs: The system of record. Scheduling, claims, basic charting.	Ours: FHIR R4 layer on top. We do not replace EHRs; we make them produce evidence-grade chronic care.
Chronic-care & CoCM platforms HealthViewX · Concert · NeuroFlow · PainScript · ChartSpan	Theirs: Pain monitoring and care-management workflow. Validates the category.	Ours: Integrated pain + BH loop with outcome-attainment forecasting and a longitudinal Care Graph compounding across patient-months.
Healthcare data infrastructure Tempus · Flatiron · ConcertAI · Komodo · Veeva	Theirs: Oncology and claims-grade datasets. The scaled comparable for what we are becoming.	Ours: The Tempus-shape company for chronic pain. Built through deployed clinical workflow capturing dense PRO + behavioral signal that claims data misses.

Why incumbents do not pivot here

For each, getting good at chronic pain + BH personalization means rebuilding their data architecture, not extending it. Hippocratic's economics depend on voice-call volume, not response-trajectory depth. Abridge's workflow value depends on encounter density, not monthly care planning. Tempus optimized for oncology multi-modal data; chronic pain is structurally different — higher PROM frequency, lower diagnostic richness, deeper behavioral integration. None of these data architectures is wrong; they are simply the wrong shape for this surface.

What we understand that they don't is two-layered: clinically, that chronic pain patients are systematically over-medicated rather than under-medicated, that BH is the rate-limiting factor on pain outcomes, and that trajectory is predictable by Week 4; structurally, that capturing this requires longitudinal multi-modal phenotype-response data inside an integrated specialty workflow — which voice agents, ambient documentation, medical search, EHRs, and care-management platforms each miss in a different way.

Why May 2026

A 12–24 month deployment-density window.

Six forces converging at once. Several of them did not exist 12 months ago.

Chronic pain remains operationally unmanaged at scale

62 million U.S. adults (24.3%, 2023). ~21 million Medicare beneficiaries. ~40% with clinically significant depression or anxiety (2025 JAMA Network Open meta-analysis). Treatment is still trial-and-error.
The post-opioid drug pipeline is arriving

FDA approved suzetrigine in January 2025 — the first first-in-class non-opioid analgesic in three decades. FDA required class-wide opioid labeling changes in July 2025. Nav1.7/1.8 inhibitors from Vertex, Lilly, Eliem are in late-stage development. Response is phenotype-dependent and behaviorally modulated — exactly the matching problem the empirical patterns address.
CMS ACCESS Model launches July 5, 2026

10-year voluntary outcome-aligned program. Chronic MSK pain and behavioral health are 2 of 4 founding tracks. CMS has accepted 150+ first-cohort organizations. Application deadline extended to May 15, 2026. ACCESS Payer Pledge from plans representing 165M+ lives for 2028 alignment. ACCESS expenditures begin entering ACO benchmark calculations in 2028.
CMS pays for monthly chronic pain & BH care

G3002 / G3003 reimburse monthly chronic pain management. CY2026 PFS expanded behavioral-health pathways through APCM add-on codes G0568/G0569/G0570. The codes are real; the workflow is hard to operationalize without integrated software.
Foundation models can now synthesize multimodal longitudinal data

Healthcare-tuned current-generation models can finally turn EHR + PROMs + medications + wearable summaries + clinician judgment into structured care plans. The capability arrived in 2024–2026. The deployment infrastructure to capture the resulting outcome data is what we are building.
FCA enforcement is at record levels

$6.8B+ FY2025 recoveries; $5.7B+ in healthcare. Weak AI billing tools are dangerous. Clinician-supervised, source-grounded, compliance-first specialty infrastructure is the defensible answer. The CDS exemption posture, paired with NIST AI RMF discipline, is now a market requirement — not a nice-to-have.

Chronic pain and behavioral health clinicians have more treatment options, more patient data, more outcome-aligned payment support, and more compliance scrutiny than ever — but no personalized care agent connects all of it. The deployment-density window before incumbents reposition is approximately 18–24 months. — XPain.AI thesis

The team

A clinician–operator pair, built for this surface.

A clinician-supervised healthcare AI startup needs ground-truth clinical workflow expertise and consumer-scale AI execution. Two founders. Two paths. They met in March 2024.

Dustin

Co-founder · Clinical · Polaris Therapy

Johns Hopkins-trained counselor (M.S. 2017). Owner-operator of Polaris Therapy LLC (Yakima, WA) — Medicare Part B-enrolled chronic-care practice. Delivered 100,000+ minutes of integrated chronic pain + behavioral health care in 2025 — the longitudinal observational caseload from which the agent's reasoning architecture was extracted.

Polaris is XPain.AI's first deployed site, becoming a paid customer at fair market value via MSA reviewed by healthcare counsel. The relationship is disclosed transparently and structurally compliant with AKS.

dustin@xpain.ai

Iskander Rakhman

Co-founder · CEO / CTO · Engineering & product

Shipped Botify AI (1M+ downloads, pre-GPT-3.5 OpenAI commercial partner) — multi-turn personalization at consumer scale before the foundation-model wave. Built ORBI to $8M ARR in 10 months — GTM execution into specialized buyer segments. Senior infrastructure engineering background; ships v1 in weeks, not quarters.

Met Dustin in March 2024. The time since has been spent translating Dustin's observed clinical patterns into the agent's reasoning architecture, and the post-opioid + ACCESS regulatory architecture into product scope.

iskander@xpain.ai

Get on the calendar

A 30-minute documentation & personalization review.

We apply the 5-pattern framework to your panel: where care plans are missing data, where medication-safety review is incomplete, where PROM capture is weak, where behavioral integration is incomplete, and where between-visit data is going unused. Most independent pain practices bill G3002 on an estimated 8–15% of their chronic pain panel; the documentable rate is closer to 40–60%. The gap is documentation, not eligibility.

Book a 30-minute review Read our compliance receipts

Or write a founder directly · iskander@xpain.ai · dustin@xpain.ai
Pharma medical affairs & investigator-initiated studies · partnerships@xpain.ai · ACCESS implementation partners · access@xpain.ai

Frequently asked

The hard questions, answered straight.

Where do the 5 patterns come from? Is this just one clinician's opinion?

The patterns were observed by Dustin across 100,000+ minutes of weekly longitudinal integrated chronic pain + BH care delivery in 2025 — a structured observational caseload, not casual clinical experience. They are the architectural patterns the agent encodes into structured decision support.

The mechanism-level claims align with the published literature on central-sensitization neurobiology (e.g., Pattern 02 — BH as rate-limiter on pain outcomes — is consistent with descending PAG/RVM serotonin/norepinephrine pathways and NMDA-mediated wind-up). The observational claims (e.g., Pattern 01 — over-medication as the dominant gap) are framed as patterns Dustin observed in his caseload, not as universal claims about all chronic pain patients.

The validation step — converting the observational architecture into structured prospective Care Graph data across 4–6 design partner practices — is the next 12 months. Pattern recognition is the starting architecture; multi-practice longitudinal data is what converts it to evidence.

Does the agent require Dustin in every loop?

No. The agent generalizes the pattern recognition into structured decision support that any QHP at any deployed site can use. Dustin's observation produced the architecture; the architecture scales to thousands of patient-months without him in the loop. Polaris remains the validation site, not the bottleneck.

This separation is the difference between "personalized medicine encoded as software" and "one clinician's manual workflow productized." The first scales; the second does not.

Polaris Therapy is Dustin's practice — isn't that self-dealing?

Polaris is Dustin's separately-owned practice and is XPain.AI's first deployed customer at fair market value, via an MSA reviewed by healthcare counsel. The arrangement is disclosed transparently in the YC application, on this page, and to every prospective customer or investor.

Flat pricing applies across all customers — Polaris included — so the structure is upstream of any Anti-Kickback Statute analysis. We do not need to invoke a specific AKS safe harbor; the relationship structure does not implicate the statute in the first place.

Is this FDA-regulated?

No — at current scope. We meet all four factors of the 21st Century Cures Act CDS exemption (21 U.S.C. § 360j(o)(1)(E)): we do not interpret raw signals; we display medical information; we support clinician recommendations; and we enable independent review of the basis for every recommendation.

If we ever added autonomous clinical decision-making, direct-to-patient diagnostic claims, raw physiological signal interpretation, or our own hardware, FDA scope would expand and a Q-Sub / De Novo / 510(k) pathway would become relevant. Each is a deliberate red line in product design.

How is this different from Abridge or Epic AI Charting?

They own encounter documentation — ambient transcripts, note drafting, order queues, broad specialty breadth. We are structurally adjacent: we operate at the level of monthly chronic care planning, not visit capture. Our buyer is the independent specialty practice, not the enterprise health system. Our data is longitudinal patient-reported phenotype-response, not ambient transcripts. We sit on top of any EHR via FHIR R4 and do not compete with them.

What about the False Claims Act?

"Human in the loop" is not enough if the human is rubber-stamping. Our architecture is built for FCA defensibility: source-grounded fact attribution; element-level QHP attestation; pre-claim review at the compliance layer; flat pricing that removes utilization incentives; audit packet ready on demand.

FY2025 FCA recoveries were $6.8B+ overall, $5.7B+ in healthcare. We architect for that environment. Pattern 05 of the empirical foundation — that documentation completeness predicts clinical continuity — is the same architectural commitment that makes the chart audit-defensible.

Do you take a percentage of claims or captured reimbursement?

No. Flat fees only — base platform, per-QHP seat, patient-panel tiers. We do not pay for referrals. We do not share claims revenue. This is upstream of Anti-Kickback Statute analysis, not a safe-harbor argument.

What outcomes data do you have?

We are honest about stage. We do not yet have prospective RCT data on the platform itself. We have four classes of evidence: (1) direct clinical observation — 100,000+ minutes of integrated pain + BH care in 2025 producing the 5 architectural patterns; (2) mechanism-based RCT evidence on the combinations the agent recommends (Gilron Lancet 2009; pregabalin + duloxetine; LDN in fibromyalgia); (3) reimbursement and capture-rate evidence from independent pain practices; (4) team execution pattern from prior products.

By Demo Day (early October 2026) we plan first 100+ structured patient-months across PEG, PHQ-9, GAD-7, sleep, function, adherence, and side-effect trajectories — converting the observational caseload into prospective Care Graph data. By Series A, 6–12 months of multi-practice longitudinal data sufficient for a peer-reviewed pilot study.

What about the ACCESS Model — what if it changes?

ACCESS is an accelerant, not the foundation. The product works through Track 1 (practice SaaS) and Track 3 (pharma medical affairs / RWE), neither of which depends on ACCESS adoption. The patient companion and the data layer remain core regardless of CMMI outcomes.

That said, ACCESS is gaining political momentum: 150+ accepted first-cohort organizations, deadline extended to May 15, 2026, broader CMMI portfolio (BALANCE, MAHA ELEVATE, GLOBE/GUARD) signaling sustained chronic-disease focus. We treat the model as a tailwind, not a dependency.

Can you integrate with my EHR?

FHIR R4 across ModMed, ChartLogic, PrognoCIS, Athena, Elation, eClinicalWorks, NextGen, AdvancedMD, and others. We do not replace your EHR. We layer on top of it and feed structured documentation back when the QHP attests.

How do you price?

Flat subscription: base platform fee ($1,500–$3,000/month), per-QHP seat ($300–$700/month), and patient-panel tiers ($20–$40 PMPM blended; lower for ACCESS-aligned cohorts). A typical 4-clinician practice with 200 active patients runs roughly $80–$130K ARR. Medicare reimbursement on the same panel exceeds these levels by a wide margin under G3002 alone.

Where can I see the agent in action?

Three options: (1) book a 30-minute documentation & personalization review — we walk through your panel using the 5-pattern framework; (2) read our compliance receipts and workbench output on this page; (3) email iskander@xpain.ai for a live demo on a synthetic patient.

The personalized care agent for chronic pain& behavioral health.