Clinical Agentic AI Platform OS· NO PHI Patent Pending Multi-Agentic AI Live in Production HIPAA-Eligible

Patient-centric, multi-agentic clinical AI. Zero LLMs in the diagnostic loop.

P.AI Global Health (Patient AI) is a coordinated system of specialized AI agents — orchestrated around a deterministic clinical core — that turns what a patient says, by voice or text in 23 Indic languages plus English, into a physician-grade SOAP note with ICD-10 codes in under three seconds. Every clinical decision traces to an auditable Python rule. Re-runs produce byte-identical output.

Run the live demo → See the architecture

What no one else has

Four engine families with zero venture-backed competition.

CARD 1

Global pediatric primary care

WHO IMCI under-5 algorithm, deterministic, in 23 languages. The first patient-facing implementation of the WHO's own pediatric primary-care framework — not an LLM approximation of it.

CARD 2

Cross-system pharmacology

The first deterministic Ayurveda + Siddha drug-herb interaction engine in the world. Rasa-shastra heavy-metal flags. Public-domain pharmacology codified as Python rules — patentable as deterministic implementations.

CARD 3

Caregiver burden

The first explicit caregiver-side engine in any clinical OS. Zarit Burden Interview operationalised at the point of intake — because the caregiver is the second patient, and no incumbent has built for them.

CARD 4

Nursing & PT instruments at home

Braden · Morse · STEADI · CAM · MUST · Bristol · NRS · ESAS · PSFS · ODI · DASH. Validated nursing and physiotherapy instruments, patient-self-administered, deterministic, longitudinally owned by the patient.

What's New · May 2026 Build

57 clinical engines. 626 tests passing. 28/28 validator-bank cases passing.

Since the original 26-engine submission, the platform has expanded into 31 additional clinical-depth, infectious-disease, accessibility, global pediatric, cross-system pharmacology, caregiver, and nursing-bundle engines — and every line is still deterministic Python with zero LLMs in the diagnostic loop. Every claim below is verifiable against the live admin panel, the public pytest suite, and the one-click clinical validator bank.

+31 ENGINES

57 deterministic clinical engines

Cardiology · Pulmonology · Mental Health depth · Hepatology · Hematology · Rheumatology · Women's & Men's depth · Dermatology · Ophthalmology (emergent + outpatient) · ENT (emergent + outpatient) · Toxicology · Geriatrics · Oncology emergencies · General Surgery · Pediatric depth · Endocrinology · Nephrology · Acute Infectious (COVID · Influenza · RSV · Cellulitis · Necrotizing Fasciitis · Endocarditis · Osteomyelitis · C. diff · Diabetic Foot · PID · Aspiration) · Vaccine-Preventable / Emerging (Pertussis · Measles · Mumps · Varicella · HFMD · Lyme · RMSF · Rabies · Tetanus · Mpox · Mononucleosis · HSV Encephalitis) · Universal Accessibility · Disability Resources · WHO IMCI (global pediatric primary care) · Traditional Medicine cross-system pharmacology (Ayurveda + Siddha) · Caregiver Burden (Zarit) · Nursing Assessment Bundle (Braden + Morse + STEADI + CAM + MUST + Bristol + NRS + ESAS) · PT Outcomes (PSFS + ODI + DASH).

+8th LAYER

8-layer drift-proof safety stack

Abbreviation Resolver → Input Hygiene Auditor → Input Security Guard → Temporality → OLD CARTS → Negation+Demographic → Severity-Lock+Dedupe+Accessibility Reserved Lane → Input-Provenance Lock. The CRITICAL-no-drop invariant guarantees that every critical risk flag survives priority ranking.

626 / 28-28 PASS

Two independent green surfaces

626 backend pytest cases pass in ≈75 seconds. The 28-case clinical Validator Bank runs end-to-end through the live API in ≈75 seconds — 28/28 PASS. Same input. Byte-identical output. Every. Single. Time.

🟢

Anti-drift commitment. The numbers above are pulled from the admin panel and live CI runs. Not marketing claims — reproducible facts. Independent clinical validators run the full bank → · One-click acute ACS demo →

🆕

May 22, 2026 — Engines 101–105 shipped. WHO IMCI (global pediatric primary care), Traditional Medicine cross-system pharmacology (Ayurveda + Siddha drug-herb interaction matrix), Caregiver Burden (Zarit), Nursing Assessment Bundle (Braden + Morse + STEADI + CAM + MUST + Bristol + NRS + ESAS), and PT Outcomes (PSFS + ODI + DASH). +100 new tests. Zero regression.

AI Terminology · Plain English

Why P.AI Global Health (Patient AI) doesn't hallucinate.

Four short definitions, written so a patient, an investor, or a regulator can read them in two minutes.

What is "hallucination" in clinical AI?

When a large language model (LLM) generates text, it predicts the next word based on patterns in its training data. It does not consult a fixed knowledge base. When it does not know an answer, it invents one that sounds plausible. This is called hallucination — fabricated content delivered with the same fluent confidence as accurate content.

Why is this dangerous in a clinical setting?

NBC News recently reported on OpenEvidence, a $12 billion doctor-side LLM literature-search tool. Their own CEO publicly admitted: "We're like 1 percent of the market in apps that think about hallucination as a limitation. 99 percent is still up for grabs." For an unsuspecting physician trusting an AI's answer, a hallucinated guideline citation, a fabricated drug interaction, or a non-existent contraindication can directly cause a wrong diagnosis or a wrong prescription. The patient pays the cost.

What does "deterministic" mean in computing?

A deterministic system, given the same input, always produces the same output. A calculator is deterministic. An LLM is not. Deterministic computing is reproducible, auditable, and testable — three things clinical regulators require.

How P.AI Global Health is deterministic from the ground up

Every clinical risk flag, ICD-10 suggestion, and SOAP item runs through 57 deterministic rule engines written in plain Python. Same patient input → same output, every time. Every flag is traceable to a specific line of code, a clinical guideline, and the patient-text trigger that fired it. No large language model touches the clinical decision layer. LLMs are used only as bounded peripherals — translating Indic languages and structuring conversational chat — never to draft a diagnosis, never to invent a guideline, never to generate a SOAP from scratch.

🟢

The benefit. The platform cannot hallucinate on the clinical decision layer because there is nothing in that layer capable of generating new content. It does not interpret the patient's history; it assembles it. Every line of the SOAP, every flag, every ICD code is the result of a rule firing — the physician can audit any decision back to source.

For Physicians & Providers · Cures Act-Aligned

Patient voice → Structured clinical intelligence of past medical history — for physician review.

P.AI Global Health does not interpret the patient's history. It assembles it. The patient brings together records from disparate sources — primary care, specialists, ER visits, foreign hospitals, pharmacies — that have likely never lived together inside any single EMR or EHR your practice can see. Under the 21st Century Cures Act, the patient has the right to consolidate that record across providers. P.AI is where they do it.

A structured clinical brief in SOAP format physicians already know how to read

The output surfaces past clinical flags from sources your system never had access to — old allergies, prior surgical complications, medication interactions, missed screenings — so you can spend the visit on the chief complaint instead of reconstructing history.

You choose the depth

Open the view you want from a single dropdown: Physician Summary (condensed), Full SOAP (default), Past History + Clinical Notes (chronological surgical-history timeline included), or Differentials + Engine Provenance (every flag tagged with its rule_id, source_guideline, and trigger tokens).

Cures-Act-aligned provenance per flag

Every risk flag carries its rule_id, source guideline, and the patient-text trigger that fired it. A complete, auditable trail — no black box.

The physician remains the decision-maker

Nothing is invented. No large language model touches the clinical decision layer. P.AI surfaces, organizes, and flags — the physician diagnoses, prescribes, and signs.

⚠️

Sandbox notice. The live demo runs on synthetic data only. Every generated PDF carries a diagonal "SANDBOX — NOT FOR CLINICAL USE — SYNTHETIC DATA ONLY" watermark. The production deployment activates a fully HIPAA-compliant, zero-trust architecture post-funding.

Universal Accessibility · ADA · §1557 · DPDPA

The only patient-facing healthcare AI that treats accessibility as first-class clinical workflow.

1 in 4 adults lives with a disability. Every other healthcare-AI app silently fails them. P.AI's accessibility layer is plumbed end-to-end: intake UI, deterministic flag-routing, and a reserved lane in the priority filter that guarantees an accessibility request can never be hidden by clinical noise.

🤟 Deaf / Hard of Hearing

Sign-language flag for ASL · ISL · BSL · LSF · Auslan + interpreter request (in-person · VRI · phone · family). ICD-10 Z71.9. Engine 49 reserved lane.

Open intake →

👁 Low Vision & Blind

Toolbar with text-size scaling (×1.15 / ×1.30), AAA yellow-on-black high-contrast palette, native browser read-aloud. Screen-reader landmarks across every page. ICD-10 H54.7.

Open intake →

🗣 Plain Language & Low Literacy

Plain-language SOAP mode written at the 5th-grade reading level, in any of 23 Indic languages or English. Z55.9. Surfaces alongside the physician-grade note — never replaces it.

Open intake →

🧩 Cognitive / AAC Users

Pictogram & symbol-board input modes. Caregiver-assisted intake with AAC device recognition. Reserved lane ensures the accessibility signal survives all clinical aggregation.

Open intake →

🌐 Spoken Language Interpreter

23 Indic languages + English via Sarvam AI STT. Interpreter request flag for spoken-language coverage when a human interpreter is required. End-to-end SOAP translation.

♿ Floating Accessibility Toolbar

Present on every page. Text-size · high-contrast · read-aloud. Persists to localStorage. Zero clinical content blocked behind a preference gate.

🛡

Reserved-lane invariant (audited Feb 16, 2026). Accessibility flags pass through the priority filter and the downstream flag aggregator without being capped by max-flags or category limits. Verified by 10/10 dedicated regression tests. ADA-aware. §1557-aware. DPDPA-aware (India).

Engine 50 · Disability Resources & Caregiver Hub

The first healthcare AI that gives caregivers the benefits list, automatically.

When a patient or caregiver intake mentions Autism Spectrum, Down Syndrome, Cerebral Palsy, Intellectual Disability, Fragile X syndrome, Rett syndrome, or global developmental delay, P.AI surfaces a curated, region-aware box of government + nonprofit resources below the SOAP note — automatically. No portal-hunting. No call-trees. No paywall. Web-verified for 2026.

🇺🇸 United States — Federal & State

SSI / SSDI eligibility links + state-level intake pathways
Medicaid HCBS waivers (Katie Beckett · Family Support · 1915c)
IEP / 504 plan rights under IDEA · ED & OCR contacts
ABLE accounts (529-style tax-advantaged disability savings)
State developmental-disability agency directory (all 50 states)
National Down Syndrome Society · Arc · Autism Self-Advocacy Network · Autism Speaks (neutrally surfaced)

🇮🇳 India — National & State

UDID Card (Unique Disability ID) — application + benefits
RPwD Act 2016 — 21 categories of recognized disability
Niramaya Health Insurance (National Trust) — ASD · CP · MR · Multiple Disability
Sugamya Bharat Abhiyan accessibility benefits
State-level disability pension & education scholarships
ABDM-ready integration path for Ayushman Bharat Health ID

❤

For caregivers — re-included. Caregivers experience ~30% prevalence of depression. Engine 50 routes the family to peer forums, respite-care directories, and federal/state benefit applications the moment a developmental-disability condition is detected — region inferred from intake language (USA default; India surfaced when patient language is Tamil/Hindi/Telugu/Malayalam/Kannada/Marathi/Bengali/Gujarati/Punjabi/Urdu). All resources surface neutrally — no organizational endorsement.

The Problem

$1.67 trillion wasted. Five stakeholders failed at once.

U.S. healthcare loses approximately $1.67 trillion every year across six categories of waste — administrative complexity, failure of care delivery, pricing failure, fraud and abuse, low-value care, and missed prevention. Globally, the picture is worse. The system is failing every stakeholder simultaneously — and the patient, who should be at the center, is the one most invisible to it.

👤

Patients

Disempowered. Only 12% of Americans have proficient health literacy; globally the gap is wider. Lost in the maze, no second consults, intimidated by jargon written at a postgraduate reading level — usually in a language they don't speak.

⚕

Providers

Burned out. 50%+ of physician time is lost to documentation, prior authorizations, and admin work instead of patients. Burnout is the leading cause of early retirement from medicine.

💸

Payers

Overloaded. Billions wasted on duplicate imaging, exploitative billing, fraud, and overlooked drug/herbal effects (e.g., ashwagandha → creatinine rise) that trigger cascades of unnecessary diagnostics.

❤

Caregivers

Emotional and physical burnout. ~30% prevalence of depression among caregivers. Quiet, unfunded, often entirely invisible to the formal healthcare system.

💊

Pharmacists & NFIs

Underutilized. Pharmacists and Non-Formal Influencers (community health workers, ASHAs, lay counselors) are vital for patient education and adherence — yet remain locked outside the clinical loop.

🌍

The Underserved, Globally

1.4 billion Indians. Rural Africa. Spanish-speaking workers in Florida. Tamil-speaking elders. Patients with no easy access to a physician, in their own language, in their own time.

⚡

The cost of doing nothing. $1.67T in U.S. waste annually. Roughly 10% of global GDP spent on healthcare. Yet patients still fax forms, wait weeks for records, and struggle to understand jargon. Healthcare today is Goliath — massive, entrenched, expensive, resistant to change.

An Open Letter from the Founder

The patient's voice was too often absent.

For over 40 years, I worked as a pharmaceutical executive, helping to bring new therapies into the hands of physicians and patients. I saw innovation change lives — and I also saw where the system fell short. In conference rooms, we spoke of patient outcomes; yet the patient's own voice was too often absent.

Every patient lives inside two worlds: the lived experience of illness, and the fragmented system that records it. I have lived in both. As a patient navigating multiple chronic conditions, biologic therapies, and treatments that worked but were never covered — I have experienced firsthand what millions endure daily. Side effects no one warned about. Specialist visits chasing a problem a single conversation could have resolved. Records trapped in someone else's system. Loneliness inside a fluorescent waiting room.

P.AI Global Health (Patient AI) was not born from theory. It was born from that lived experience — and from a single conviction:

The patient must be at the center. Not the institution. Not the algorithm. Not the payer. The patient.

This is the gap P.AI Global Health (Patient AI) was created to fill — a patient-centric, multi-agentic AI healthcare companion that finally turns what patients say into what physicians, payers, and the underserved can actually use.

For investors and serious partners: a full Founder Open Letter — including specific clinical journeys, regulatory analysis, and a four-decade career résumé — is shared on request through the secure investor channel. Request the letter →

Why Now

The law has shifted. The AI is ready. The waste is intolerable.

For the first time, the regulatory environment, the multi-agentic AI capability, and the public urgency are aligned. The cost of patient data-blocking is now higher than the cost of patient empowerment.

📜 The Legal Mandate Has Shifted

HIPAA grants enforceable rights to access health information. The 21st Century Cures Act goes further — it prohibits "information blocking." The liability has flipped: today the risk is in failing to share patient data, not in sharing it.

⚠ Yet Patients Are Still Locked Out

Patients are still faxing forms, waiting weeks for records, and struggling with jargon-filled reports in a language they don't speak. Plain-language disclosure is now legally required — yet most documentation is written at a postgraduate reading level, in English only.

🏥 The Triage-to-Record Gap

Triage — the first critical act in emergency care — remains siloed from the medical record. A brief triage note can mean life or death, yet it rarely connects to a patient's longitudinal history. EMTALA and malpractice liability surface every time it fails.

🧠 Multi-Agentic AI Is Production-Ready

Today's AI infrastructure makes it possible to deploy a coordinated multi-agentic pipeline — specialized agents for intake, redaction, normalization, longitudinal memory, drafting, and guardrails — without ever surrendering clinical decisioning to a black box. P.AI Global Health (Patient AI) is built exactly this way.

Patient Journeys

Real categories of failure. Real waste prevented.

Each scenario below is a real category of failure inside U.S. and global healthcare today. Each is something a multi-agentic AI companion, deployed with consent and clinician oversight, would catch — saving lives, money, and dignity.

DRUG–HERBAL INTERACTION

The ashwagandha–creatinine cascade

An elderly patient self-medicates with ashwagandha. Creatinine rises. Multiple specialist visits and tests are ordered. P.AI Global Health (Patient AI) detects the herbal–lab link early, prevents payer waste, and symptoms resolve once the supplement is stopped.

BIOLOGIC SIDE EFFECT

The Dupixent–psoriatic-arthritis link

A patient on Dupixent develops new joint symptoms. P.AI Global Health (Patient AI) connects the symptoms with the biologic and escalates the case for provider review — surfacing a known but commonly overlooked association.

BILLING FRAUD

The duplicate MRI

A patient is billed twice for the same MRI. P.AI Global Health (Patient AI) flags the duplicate to both the patient and the payer — protecting the patient's wallet and the payer's solvency simultaneously.

CAREGIVER MENTAL HEALTH

The depressed caregiver

A caregiver reports symptoms of depression to P.AI's 24/7 companion. The system routes them to peer support, clinical resources, and — where available — philanthropic mental-health programs.

LANGUAGE EQUITY · USA

The Spanish-speaking Florida worker

A construction worker with chest discomfort uses voice input in Spanish. P.AI Global Health (Patient AI) produces a structured, English SOAP draft for the nearest ER — avoiding an interpreter delay that could cost minutes during a cardiac event.

ACCESS EQUITY · INDIA

The mother in rural Tamil Nadu

A mother describes her child's fever in Tamil. P.AI Global Health (Patient AI) produces a structured pediatric intake for a remote physician — preserving the patient's language, the clinician's time, and the child's life.

COMPANIONSHIP & LONELINESS

The elderly Tamil widow

An elderly widow uses P.AI's companion to listen to Tamil audiobooks and devotionals — easing loneliness and reducing stress that is itself a documented driver of cardiovascular and cognitive decline.

AFFORDABILITY GAP

The unaffordable cream

A patient responds well to a topical biologic that simply isn't covered by their plan. P.AI Global Health (Patient AI) routes them to philanthropic and patient-assistance programs through the Health Mart commerce layer — without selling their data.

Our Vision

Patient-centric. Causal. Multi-agentic. Global.

P.AI Global Health (Patient AI) (Patient AI) is a patient-centric, global, causal, multi-agentic AI healthcare infrastructure solution.

Our focus is not chat. It is turning patient-owned data into clinician-ready, interoperable outputs that integrate safely into healthcare workflows over time.

P.AI Global Health (Patient AI) is not a top-down, institution-centric AI product. It is a bottom-up system designed around a minimalist principle: reduce systemic bloat by empowering patients to own their health data while reducing administrative burden for clinicians.

The Solution

A patient-centric, multi-agentic AI healthcare companion.

P.AI Global Health (Patient AI) is not a chatbot. It is a coordinated system of specialized AI agents orchestrated around a deterministic clinical core. Each agent does one thing well. Every decision is traceable. The patient is at the center — and every stakeholder benefits.

👤 For Patients — Rights Made Real

HIPAA and Cures Act compliance is built-in. When patients request their data, it flows — instantly, digitally, in any language. The Patient Portable Health Vault travels with the patient, not with the hospital.

⚕ For Providers — Burden Reduced

The Ambient Scribe agent transforms triage and visit notes into structured, interoperable EMR entries. Physicians get their time back — from documentation to patients. EMTALA and malpractice liability fall because triage notes are timestamped, standardized, and linked.

💸 For Payers — Waste Recovered

The Guardrails agent surfaces duplicate testing, exploitative billing, and overlooked drug/herbal effects. Cascades of unnecessary diagnostics are prevented before they're ordered — moving billions back into the system.

❤ For Caregivers & Pharmacists — Re-Included

The 24/7 Companion agent supports caregivers with peer forums and mental-health resources. Pharmacists and NFIs are looped into adherence and patient education — finally treated as the clinical assets they are.

🌍 For the Underserved — Equity by Default

23 Indic languages plus English. Plain-language explanations. Multimodal input (voice, text, document, image). Free forever for patients. Built for villages, not just hospitals.

🛡 For Regulators — Auditable by Architecture

Every agent's output is logged. Every risk flag traces to a line of Python. Every encounter is byte-reproducible. TEFCA-aligned. FHIR-native. ABDM-ready (India). The architecture FDA and ONC can certify.

The Product

Three integrated products. One coherent vision.

Patient-first, clinician-guided, built with guardrails. Free for patients and physicians forever. Monetization happens downstream — payer / employer waste recovery and a separately-operated marketplace entity.

LIVE

P.AI Global Health (Patient AI) Clinical Engine

57 deterministic clinical engines · 8 safety layers · zero LLMs in diagnosis. Produces a SOAP note with inline ICD-10-CM codes and USPSTF Grade A/B preventive recommendations in 2.1 seconds. Live at app.paiglobalhealth.com.

BETA

MHR — My Health Records

Free lifelong patient-owned mobile health vault. Reads/writes FHIR. Companion to the clinical engine. TestFlight ready. 44/44 backend tests passing.

PLANNED

P.AI Health Mart

The patient-trust commerce layer. Exclusively affiliated with P.AI Global Health (Patient AI). Curated. HIPAA-compliant. No ads, no scams, no fraud. Read more ↓

⚡

The moat in one sentence. Every other healthcare AI is wrapping ChatGPT and praying it doesn't hallucinate a diagnosis. We took the opposite path — 57 hand-coded clinical engines, 8 safety layers, every risk flag traceable to a line of Python. Same input always produces the same output. See for yourself.

P.AI Health Mart

The patient-trust commerce layer.

P.AI Health Mart is exclusively affiliated with P.AI Global Health (Patient AI). It is the premium, curated services layer designed for one purpose: the utmost trust of the patient. No ads. No scams. No fraud. No opaque steering. No data selling.

What it will offer

Second-opinion consultations from vetted physicians
Home health-care services
Curated DME (Durable Medical Equipment) vendors
Direct-to-Consumer (DTC) wellness products
Categories will expand as deemed clinically appropriate

How trust is engineered

Every vendor curated & vetted
HIPAA-compliant end-to-end
Patient consent gates every transaction
Separately-operated LLC — firewalled by the consent-token issuer
Zero PHI crosses the commerce boundary

💡

Revenue sustainability without compromise. Health Mart enables a premium tier without ads, without data selling, and without opaque incentives. Revenue supports platform sustainability while preserving the clinical and consent boundaries that define P.AI Global Health (Patient AI).

How It Works

Four steps for patients. Seven agents under the hood.

A patient describes how they feel. The multi-agentic pipeline does the rest — visibly, traceably, with a physician always in control.

For Patients

Describe Symptoms

Voice or text. Any of 23 Indic languages or English.

AI Structures Data

The multi-agentic pipeline organizes your story clinically.

Doctor Reviews

Physician receives a structured SOAP note with ICD-10 codes.

You Own Your Data

Your record stays in your personal Patient Portable Health Vault.

Under the Hood — The Multi-Agentic Pipeline

P.AI Global Health (Patient AI) uses a deterministic, auditable agent-style pipeline — not uncontrolled autonomous reasoning. Each step is visible, traceable, and designed for physician oversight.

01 · Intake Agent

Validates, structures, and organizes raw patient input into a standardized format.

02 · Redactor Agent

Scans for and redacts potential personally identifiable information (PII) patterns.

03 · Normalizer Agent

Converts medical abbreviations and shorthand into standard clinical terminology.

04 · Longitudinal Memory Agent

Loads prior encounter history to provide clinical context continuity.

05 · Draft Clinical Note Agent

Generates a structured SOAP note with conservative clinical reasoning.

06 · Guardrails Agent

Checks for definitive diagnosis language, hallucinated vitals, and safety concerns.

🛡

The 7th agent is the physician. No agent in this pipeline publishes, prescribes, or finalizes. Every output is a draft. The licensed clinician remains the final decisioning authority — always.

Patient Portable Health Vault (PPHV)

Your health record. Owned by you. Travels with you.

Free forever for patients. P.AI Global Health (Patient AI) provides every individual with a Patient Portable Health Vault — a secure personal health record that belongs to you, not to the hospital.

🔐 Like Your Bank Vault

Your health vault is secure and under your control — similar to how you control access to your bank account. Diagnoses, medications, lab results, imaging, and full history — organized in one place that you own.

🌐 Portable Worldwide

Your health history travels with you — not trapped in hospital systems, EHRs, or EMRs. Share it with any physician, clinic, or hospital, anywhere in the world, in seconds.

🎙 Multimodal Intake

Add data by text, conversation, document upload, or lab-report scanning. The system reconciles new entries against your history automatically.

🗣 Multilingual Access

Interact in your own language — 23 Indic languages plus English at launch, more on the way. Plain-language explanations of every clinical term.

⚕

For physicians: Patient voice → Structured clinical intelligence → Draft SOAP note. P.AI Global Health (Patient AI) transforms patient intake into structured clinical summaries ready for physician review. Physicians remain fully in control of the final clinical decision and documentation.

Free Tier & Sustainability Model

Free forever for those who need it. Revenue from those who can fund it.

P.AI's revenue model is deliberately downstream from the patient. Patients are never the product. Their data is never sold. Their diagnosis is never paywalled.

FREE FOREVER

Patients

Unlimited multi-agentic intake
Patient Portable Health Vault
23 Indic + English support
24/7 companion & forums
Symptom checker & escalation

FREE FOREVER

Individual Physicians

Unlimited ambient scribe
SOAP draft + ICD-10 inline
USPSTF preventive matching
Multilingual patient handoff
Documentation-burden reduction

PAID · DOWNSTREAM

Health Mart, Payers & Partners

B2C Health Mart commissions
B2B vendor subscriptions
Payer waste-recovery analytics
Employer wellness reporting
Philanthropic NGO programs

🚫

What we will never do. Sell patient data. Run ads on the patient experience. Paywall a diagnosis. Use patient text or voice to train external LLMs. These commitments are architectural, not promotional.

The Architecture

Deterministic core. AI peripherals. Strict boundaries.

The clinical decisioning runs in pure Python — auditable, reproducible, regulatory-defensible. AI is bounded to peripheral roles: voice transcription, multimodal image input, conversation formatting, translation. AI never touches diagnosis.

        PATIENT / PHYSICIAN / MHR APP — 23 Indic + English
                           |
              [ Cloud Load Balancer + WAF ]
                           |
        +------------------+------------------+
        |                                     |
   FRONTEND (React)                   BACKEND (FastAPI)
   Firebase Hosting                      Cloud Run
                                              |
       +--------------------------------------+--------------------------------+
       |                                      |                                |
DETERMINISTIC ENGINE            VERTEX AI (peripherals)          CLOUD HEALTHCARE API
57 clinical engines              Claude scribe                       FHIR store
8 safety layers                  Gemini patient chat                 ABDM-ready
Layer 0.5 Input Hygiene Auditor  Gemini Vision (rash, labs)          MHR reads/writes
Layer 6 Input-Provenance Lock    Cloud TTS · 23 languages            here
ICD-10 + USPSTF inline           Cloud STT (Sarvam fallback)
ZERO LLM in diagnosis            Never touches risk flags
       |                                                                       |
       |               +------------------------------------+                  |
       |               CONSENT-TOKEN ISSUER (Cloud Run)                 |
       |               → category + zip only, ZERO PHI                 |
       |               → HEALTH MART (separate LLC)                    |
       |                                                                       |
       +-----------------------------+-----------------------------------------+
                                     |
            DATA LAYER — HIPAA-eligible · BAA in place
            MongoDB Atlas · Cloud Storage · BigQuery analytics
                                     |
            COMPLIANCE — Cloud Logging · KMS · IAM · Cloud DLP

🟢

What is Deterministic

The 57 clinical engines
Risk flag severity locking
ICD-10-CM assignment
USPSTF preventive matching
Negation, temporality, OLD CARTS

🟡

What is AI-Assisted

Voice-to-text (Sarvam + Cloud STT)
Multimodal vision (Gemini)
Ambient scribe (Claude)
Patient chat (Gemini)
SOAP translation (23 langs)

🟣

What is Firewalled

Consent-token issuer
Health Mart commerce (separate LLC)
No PHI crosses the boundary
Audit trail in Cloud Logging
Patient consent gates everything

Trust & Guardrails

Safety is a system design choice, not a marketing claim.

P.AI Global Health (Patient AI) is built with explicit boundaries that make unsafe behavior architecturally impossible — not just unlikely.

🚫 No Autonomous Diagnosis

AI outputs are explicitly labeled "Draft." All risk flags require licensed clinician review and approval before any action is taken.

👁 Human-in-the-Loop Always

Physicians remain in control. No automated publishing, no automated prescribing, no automated insurance claims.

🔒 Input-Provenance Lock (Layer 6)

The engine can only act on words the patient actually said. The system cannot infer, fabricate, or silently expand clinical data. Zero hallucination by architecture.

📜 Auditability by Default

Every pipeline step is logged with timestamps. Every encounter is byte-reproducible. A regulator can reconstruct exactly why any risk flag fired.

🛡 HIPAA-Eligible Architecture

Encryption at rest (AES-256) and in transit (TLS 1.3). Role-based access controls. BAAs in place with all infrastructure providers. SOC 2 alignment planned.

⚠ Emergency Awareness

High-risk symptom patterns surface visible escalation warnings. The platform clearly directs users to call emergency services when appropriate. Not a 911 replacement.

⚠

What We Explicitly Do NOT Do. No autonomous diagnosis. No treatment recommendations. No bypassing clinician review. No emergency services. No sale or monetization of patient data. No opaque or "black-box" clinical outputs. For medical emergencies, call 911 or your local emergency number.

Roadmap

From MVP to Google Cloud — five disciplined phases.

Currently in MVP phase on managed Kubernetes. Production migration to Google Cloud is planned upon funding and Scale Tier approval. Each phase is independently deployable.

Phase 0 · Foundation · Now

GCP project setup & HIPAA BAA

Activate Google Cloud project. Sign HIPAA BAA. Configure IAM. Enable Vertex AI, Cloud Run, Cloud Healthcare API, Cloud Storage. $0 cost.

Phase 1 · Lift & Shift

Migrate MVP to Google Cloud Run + Firebase Hosting

Containerize the FastAPI backend on Cloud Run. Move React frontend to Firebase Hosting. Migrate MongoDB to MongoDB Atlas on GCP. No code changes — same MVP, GCP-hosted.

Phase 2 · LLM Peripherals on Vertex AI

Claude, Gemini, Vision, TTS, STT on Vertex AI

Move ambient scribe and patient chat to Vertex AI Model Garden. Add Gemini Vision for multimodal patient input. Add Google Cloud TTS for 23-language voice synthesis. Deterministic engine remains untouched.

Phase 3 · FHIR + MHR Integration

Cloud Healthcare API as FHIR-native data layer

Provision Cloud Healthcare API FHIR store. Closed loop between MHR app, P.AI Global Health (Patient AI) clinical engine, and longitudinal patient records. ABDM-ready (India). Build Consent-Token Issuer for the Tier 2 commerce firewall.

Phase 4 · Analytics + Payer Reporting

BigQuery for waste-recovery monetization

De-identified analytics pipeline into BigQuery. Generate payer-ready and employer-ready reports on avoided ER visits, prevented unnecessary tests, and ICD-10 frequency.

Phase 5 · Antigravity + CI/CD

Modern dev workflow

Adopt Google Antigravity for daily development. Cloud Build for CI/CD. Vertex AI Pipelines for automatic regression test runs on every push.

Estimated annual GCP run cost after Phase 5: ~$3K–$5K. Covered comfortably by Google for Startups Scale Tier credits, with multi-year headroom.

Global Impact — United States & India

Designed for advanced and emerging healthcare systems.

P.AI Global Health (Patient AI) is designed as a global patient-centric platform addressing healthcare challenges in both advanced and emerging healthcare systems.

🇺🇸 United States

Physicians face increasing administrative burden and fragmented medical records across multiple healthcare systems. P.AI Global Health (Patient AI) helps organize patient-reported data into structured, clinician-ready documentation, reducing documentation workload while improving care continuity.

🇮🇳 India

India faces a different challenge: large population, multilingual environments, and limited physician availability in many regions. AI-assisted intake and multilingual health interaction can significantly expand access to care while allowing physicians to focus on higher-value clinical decision making.

🌐

A shared opportunity. By combining multilingual patient input, AI-driven structuring of health information, and physician-in-the-loop oversight, P.AI Global Health (Patient AI) aims to create a scalable health infrastructure layer capable of supporting both developed and emerging healthcare ecosystems.

Why This Matters

One pebble at a time.

The Crow and the Pitcher — an Aesop fable

In an old story, a crow finds water it cannot reach. Instead of forcing the problem, it adds small pebbles — raising the water one step at a time.

Healthcare works the same way. Your story matters — but it often gets lost or fragmented.

P.AI Global Health (Patient AI) helps structure your information so your doctor can see clearly from the start.

— The founder's second proverb, alongside the Tamil one above.

Team

Two-generation founding intent.

P.AI Global Health (Patient AI) is founder-funded and founder-built. Clinical and technical guidance is provided by informal advisors who are not employees, officers, or directors of P.AI Global Health (Patient AI) LLC.

Founder & CEO

Sadasivam "Sivam" Eniasivam

Former Pharmaceutical Executive · 40+ years

40+ years of regulated-healthcare industry leadership across Glaxo, Novartis, and Knoll — spanning regulatory, commercial, and clinical operations. Long-time patient advocate focused on health equity, trust architecture, and patient-controlled longitudinal records. P.AI Global Health (Patient AI) was born not from theory but from lived experience inside healthcare systems where documentation burden, fragmented records, and unequal access affect real people every day.

Informal Advisors

Clinical Advisor — Hospital Medicine

Dr. Archana Eniasivam, MD

Associate Professor · Internal Medicine & Pediatrics

Recipient of the Society of Hospital Medicine Award of Excellence for DEI Leadership. National voice on health equity in academic medicine.

Advisory input only in personal capacity. No formal employment, governance, or operational role with P.AI Global Health (Patient AI).

Technical Advisor — Enterprise Architecture

Mr. Vish Narayan

IBM Distinguished Engineer · CTO, Industrial Sector Analytics & Optimization

30+ years of enterprise architecture, hybrid cloud, and AI optimization. Industry 4.0 / MES 3.0 consortium member.

Advisory input only in personal capacity. Not representing IBM.

Independent Clinical Validator

Dr. V. N. Devarajan, M.D.

Internal Medicine + Nephrology · 50+ years clinical practice

Former Medical Director, DaVita Dialysis Center (Luling, LA). Former Chief of Staff and ICU Co-Director, St. Charles Hospital. Has independently reviewed P.AI's deterministic clinical engine output end-to-end and confirmed clinical appropriateness for inputs given.

Independent clinical review only. Not an employee of P.AI Global Health (Patient AI).

Potential Clinical Pilot Pathway

Dr. Balaji Singh, M.S. (Surgery)

Dean, General Surgery · Professor · Sri Ramachandra Medical College, Chennai, India

Potential introduction for a structured clinical pilot in a private physician practice setting.

No formal affiliation at this time.

Intellectual Property

Patent Pending.

Notice

P.AI Global Health (Patient AI) is a patient-centric, deterministic clinical AI infrastructure platform that transforms patient-reported health information into structured clinical intelligence while maintaining physician-in-the-loop oversight.

The platform architecture, workflow design, and system concepts — including the deterministic engine framework, the 8-layer drift-proof safety stack, the patient-owned longitudinal health-data model, and the consent-token boundary architecture — represent original work developed for the P.AI Global Health (Patient AI) platform and are subject to pending intellectual property protection.

P.AI Global Health (Patient AI) LLC is incorporated in Florida, USA. Domain ownership and IP rights belong solely to the LLC and its founder. We are not affiliated with, owned by, endorsed by, or operated by any third-party domain or organization with similar naming.

Resources

Materials for investors, clinicians & partners.

Detailed materials are shared on request. For preliminary review, the live demo and public architecture overview are open to anyone.

▶

Live Demo

Run the deterministic clinical pipeline in your browser. No login required.

app.paiglobalhealth.com →

📐

Architecture Overview

Public architecture diagram and the 5-phase Google Cloud migration roadmap.

View architecture ↑

📩

Investor & Partner Materials

Competitive matrix, IP filings, financial projections, and architecture deep-dive are shared via secure data room on request.

Request access →

✉

Founder Open Letter (Private)

Full open letter — including specific clinical journeys, regulatory analysis, and a four-decade career résumé — shared via secure channel with serious investors and partners.

Request the letter →