Independence OS
Ravinder SyalFounder and CEO
Multi-time founder with 30 years in healthcare operations; scaled ventures before Independence OS and now builds AI to keep physicians independent.
30y health opsmulti-time founderHBS Exec Ed
Creating the AI ecosystem for healthcare
Building machines for human flourishing.
Independence OS is building a suite of AI products for the practice, across admin, diagnosis, insurance, and personalized care. We aim to make personalised care accessible and affordable.
Houston and BangaloreClinics in Texas and CaliforniaLaunching v1
One operating system
The system
A morning at the practice, handled.
One system across every surface of the clinic, on an ordinary morning.
Front desk Clinics
8:30Well-child visitconfirmed
9:00New patientintake sent
9:30Follow-uprescheduled
Notes Scribe
Visit summary drafted from the conversation.
ICD-10 · J06.9CPT · 99213
Eligibility Credentialing
Coverage verified
Aetnaactive
Claims RCM
Claim 8842paid
Claim 8839appealed
Supply Supply chain
Reorder placed
Flu vaccine200 doses
Quality Guardian
Well-child visits in the first 15 months (W15)82%
Every surface reads and writes to Foundry, one shared record.
The models behind the agents are ours.
Our clinics
Built in real clinics, not in a demo.
Most of a practice runs on work that has nothing to do with care: scheduling, coding, claims, follow-ups. We build the AI that takes it off the staff, proven on real patients and real billing in our own clinics before it goes anywhere else.
We own and operate a growing network of clinics across Texas and California, starting with our first, a pediatric practice in Pasadena, Texas. They are where everything we build is used first, and the bar it has to clear.
How it gets better
Run it in our clinics
Every agent ships first to a practice we operate.
Watch real visits and claims
We measure it against live work, not a benchmark.
Fix what breaks
Harnesses turn the failures into the next training set.
Extend to other practices
What clears our bar goes to everyone else's.
Practices we operate
Nag Pediatrics FlagshipPasadena, TX
Silverlake PediatricsPearland, TX
Pediatric Medical CenterPasadena, TX
Pediatric Clinic of La PorteLa Porte, TX
Babies 2 Teens PediatricsGold River, CA
Fairmont PediatricsPediatrics
Fairmont Family PracticeFamily practice
Corpus Family PracticeFamily practice
Armand PediatricsPediatrics
Nag Mindful TherapyBehavioral health
How it grows
One system, growing outward.
01
It starts in the clinic
We run our own practices. Every product is built and proven on real visits, real billing, and real patients before it goes anywhere else.
02
Everything compounds
The agents share one record and one set of models. Each visit, claim, and follow-up makes the whole system a little better.
03
Then it grows outward
From the front desk and the claim today, to patients and other practices next. One connected ecosystem, expanding from the clinic out.
The ecosystem
Everything runs around one practice.
Our clinics sit at the center. Every group of AI products plugs into the same record, from the visit and the front desk to billing and what comes next. As each one proves out in our own practices, the circle widens to others.
What we build
One system, several products.
The same products, working together on one shared foundation.
01
Foundry
One record, shared.
Every product reads from and writes to one patient record.
Foundry
Live ingest 21 /s
writescribe.noteok 24ms
syncclinics.apptok 12ms
indexrcm.claimok 18ms
Sources
Clinicslive
Scribelive
RCMlive
02
Models
Built for medicine.
Our own speech and clinical models, tuned for the exam room.
Voice model
recording
Transcribing patient reports a sore throat for three days, no fever.
drafting note
Models
Speech to textlive
Clinical NLPlive
Codinglive
03
Scribe
The note writes itself.
Record the visit. Finish the chart. Go home.
Visit note
Maria Reyes · today
ClinicianHow long has the throat been sore?
PatientAbout three days, no fever.
SSore throat, three days. No fever.
OThroat erythematous. Lungs clear.
AAcute pharyngitis. suggested Dx
PSupportive care.
ICD-10 · J06.9CPT · 99213auto-coded
04
Clinics
It always picks up.
Calls answered, visits booked, no one left on hold.
UPatient
AAgent
Transcript
AgentHow can I help?
Call
Appointmentbooked · Thu 9:30
Hold time0s
05
Guardian
Nothing slips.
Every care gap surfaced, and closed on time.
Gprocessing
Guardian
deterministic brain
HEDIS W1582%
Gaps open 24HCC suspects 3
Directives
Bishopapp reminder · Maria Reyes
Clinicscall worklist · 3 due
gapHEDIS W15 · detected
06
RCM
Every claim, clean.
Scrubbed, coded, and submitted before a denial becomes a conversation.
Revenue cycle
Coded
Scrubbed
Submitted
Adjudicated
Paid
$48.2k
Collected
7
Appeals won
Claims
Claim 8842Paid
Claim 8839 CO-16Appeal drafted
Claim 8851Submitted
07
Credentialing
The paperwork fills itself.
Eligibility and coverage, checked before the visit.
Eligibility check
Patient
CoverageActive
Ae
Aetna
Open PPO
Copay$25
Deductible$500 / $1,500
Member IDW2049 8841
Coinsurance20%
EffectiveJan 1, 2026
08
Supply chain
Never runs out.
Inventory watched, reorders placed before the shelf is empty.
Inventory analytics
142
Tracked
3
Alerts
98%
On-time
Usage · last 8h
Par levels
Gloves, mediumreorder placed
Amoxicillinok
reorder placed · gloves
09
Castle
Independence, at scale.
We buy and run the practice, so you can keep practicing.
Pediatric practice
$1.2MPasadena, TX
~5,000 patients2 providers1.8x
Family medicine
$2.1MAustin, TX
7,200 patients3 providers2.1x
Pediatrics
$0.9MKaty, TX
3,400 patients2 providers1.6x
10
Bishop v2
Care between visits.
A health assistant in every patient's pocket.
MRvoice-first
Maria Reyes
Age 52 · MRN 04821
Health Pace · biological age↓ 2.8 vs actual
49.2yrs
Patient World Model
Trust level0.78
Care plan2 open · 1 due
Medications3 active
Channelsapp · SMS · voice
SMS reminder · opened
Where we are
Launching v1 now.
v1 is going into our own clinics now. The data it produces trains the next generation of agents and models.
01
v1, now
Live in our clinics
Into our own practices, on real visits, claims, and data from day one.
02
Next
Data and models
That data becomes harnesses, stronger agents, and our own foundational models.
03
v2
Bishop and beyond
The next generation, including Bishop, and expansion beyond our own clinics.
We are growing quickly.
A physician should be able to stay independent, and still keep their attention on patients.
Running a practice has become so heavy that many doctors give up their independence just to cope. That operational weight is exactly what software should carry, so a practice can stay its own and keep care at the center.
Products
Everything a practice runs on, in one system.
Ten products on one shared foundation, grouped by where they work. Each goes to work in our own clinics first. Here is what each one does, and what it looks like.
The system
One operating system, end to end.
The layers
How they connect
One operating system, end to end. Every agent connects through Guardian, the deterministic brain that holds the rules and can veto a step, so the parts stay in sync on one shared record without wiring into each other.
The map
Every part of a clinic, and what runs it.
Foundation
01Foundrythe shared record under every product
02Modelsclinical and voice intelligence
In the clinic
03Scribethe visit, notes, and coding
04Clinicsthe front desk, calls, and scheduling
05Guardiancare gaps, HEDIS, and quality
Revenue and operations
06RCMclaims, denials, and appeals
07Credentialingeligibility and coverage
08Supply chainsupplies and inventory
Beyond the clinic
09Castlebuying and running practices
10Bishopthe patient, between visits
Foundation
Foundry
The shared record under everything.
Foundry takes in everything the clinic and the agents produce, demographics, visit notes, claims, coverage, supplies, and structures it into one record. Every other product reads from it and writes back to it, so nothing lives in a silo and nothing is entered twice.
- Unifies records across every product
- One source of truth the agents share
- Keeps clinical data clean and current
Foundry
4,812
Patients
1.2M
Records
99.9%
Synced
Live ingest 21 /s
writescribe.noteok 24ms
syncclinics.apptok 12ms
indexrcm.claimok 18ms
Sources
Clinicslive
Scribelive
RCMlive
Credentialinglive
Labssyncing
Unified record
MR
Maria Reyes
MRN 04821 · merged from 5 sources
Foundation
Models
Clinical and voice models, built in house.
We train our own models for the tasks a practice runs on, speech, clinical language, and coding. They power Scribe and our voice agents today. Foundational healthcare models, trained on what we learn in our clinics, are next.
- Speech to text tuned for the exam room
- Clinical language for notes and coding
- Foundational models on the roadmap
Voice model
recording
Transcribing patient reports a sore throat for three days, no fever.
drafting note
In the clinic
Scribe
The visit, written for you.
Scribe listens to the visit and does the documentation. Before the patient arrives it surfaces their history and the reason for the visit. During the visit it transcribes. After, it drafts the SOAP note and a plain summary, and assigns the ICD and CPT codes for billing.
- Drafts the SOAP note and summary
- Assigns ICD and CPT codes
- Surfaces history before the visit
Visit note
Maria Reyes · today
ClinicianHow long has the throat been sore?
PatientAbout three days, no fever.
SSore throat, three days. No fever.
OThroat erythematous. Lungs clear.
AAcute pharyngitis. suggested Dx
PSupportive care. Return if worse.
ICD-10 · J06.9CPT · 99213auto-coded
In the clinic
Clinics
A front desk that answers every call.
Clinics is a voice agent for the front office. It picks up the phone, books and reschedules appointments, answers routine questions, and follows up with patients, in natural conversation, without putting anyone on hold.
- Answers and places calls
- Books and moves appointments
- Handles routine follow-ups
UUser
AAgent
Transcript
AgentI have Thursday at 9:30 or Friday at 11.
UserThursday 9:30 works.
AgentBooked for Thursday at 9:30.
In the clinic
Guardian
Care that does not fall through the cracks.
Guardian watches a patient's care over time. It reads from Clinics, Scribe, and RCM to track preventive care and HEDIS measures, check dosages and adherence, and flag when someone is due for a follow-up. It is the reasoning layer behind Bishop.
- Tracks HEDIS and preventive care
- Checks dosage and adherence
- Flags overdue follow-ups
Gprocessing
Guardian
deterministic brain
HEDIS W1582%
Gaps open 24HCC suspects 3
Directives
Bishopapp reminder · Maria Reyes
Clinicscall worklist · 3 due
gapHEDIS W15 · detected
Revenue and operations
RCM
Get paid for the work you do.
RCM runs the revenue cycle. It prepares and submits claims, tracks them, and when a claim is denied it works the denial and files the appeal, so the practice collects what it is owed without the manual chase.
- Submits and tracks claims
- Works denials and appeals
- Keeps the revenue cycle moving
Revenue cycle
Coded
Scrubbed
Submitted
Adjudicated
Paid
$48.2k
Collected
7
Appeals won
Claims
Claim 8842Paid
Claim 8839 CO-16Appeal drafted
Claim 8851Submitted
Revenue and operations
Credentialing
Coverage checked before the visit.
Credentialing verifies a patient's eligibility and benefits with their payer, so coverage is confirmed before the visit and claims do not bounce for the wrong plan or a lapsed policy.
- Verifies eligibility and benefits
- Confirms coverage before the visit
- Cuts denials from bad plan data
Eligibility check
Patient
CoverageActive
Ae
Aetna
Open PPO
Member IDW2049 8841
Copay$25
Deductible$500 / $1,500
Coinsurance20%
EffectiveJan 1, 2026
Revenue and operations
Supply chain
The clinic never runs out.
Supply chain tracks what a clinic has on hand and what it is using, places orders before things run low, and predicts demand so the shelves and the fridge are always stocked without overbuying.
- Tracks inventory in real time
- Reorders before stockouts
- Predicts what a clinic will need
Inventory analytics
142
Items tracked
3
Reorder alerts
98%
On-time
Usage · last 8h
Par levels
Flu vaccine200
Gloves, mediumreorder placed
Amoxicillinok
reorder placed · gloves
Beyond the clinic
Castle
A marketplace for medical practices.
Castle is where practices are bought and sold. It is how we find and acquire the clinics we operate, and how other buyers and sellers find each other. It brings listings, diligence, and deal flow into one place.
- Lists practices for sale
- How we acquire our clinics
- Brings diligence into one place
Pediatric practice
$1.2MPasadena, TX
~5,000 patients2 providers1.8x
Family medicine
$2.1MAustin, TX
7,200 patients3 providers2.1x
Pediatrics
$0.9MKaty, TX
3,400 patients2 providers1.6x
Beyond the clinic
Bishop v2
A health assistant in every patient's pocket.
Bishop is a personalised assistant for patients. It connects to their records, labs, and day to day activity, keeps up with their care plan and follow-ups, answers questions in plain language, and nudges them at the right time. Guardian is the reasoning behind it. It arrives with v2.
- Connects records, labs, and activity
- Keeps up with care and follow-ups
- Answers in plain language
MRvoice-first
Maria Reyes
Age 52 · MRN 04821
Health Pace · biological age↓ 2.8 vs actual
49.2yrs
Patient World Model
Trust level0.78
Care plan2 open · 1 due
Medications3 active
Channelsapp · SMS · voice
SMS reminder · opened
One system, proven where it has to work first.
Every product here runs in our own clinics before it goes anywhere else. If you are a practice or a builder who wants in, we would like to hear from you.
Approach
An ecosystem of AI, built from inside the clinic.
Most healthcare software is sold from the outside in. We do the opposite. We run our own clinics, build the AI that runs them, and widen the circle from there, one product at a time.
Thesis
A machine for human flourishing: making easy things easy and hard things possible, with limitless human agency as both the means and the end.
We are building a machine for human flourishing, and we are building it on a single inversion that the rest of healthcare has backwards. A hospital survives on sickness: its fixed capital has to be fed, so its economics pull toward the bed, the procedure, the costliest 5% who consume half the spend. The clinic is the one unit whose survival depends on the opposite: keeping people well and out of the hospital.
So we put a recursive learning loop at its core, one that gets smarter with every patient it sees. It gives the patient agency over their own health, the provider a 360° view, the payer accountability, and it compresses the cost of care until the right care becomes the cheapest care. That is the whole thesis.
When doing right by the patient is also the rational economic act, prevention stops being charity and becomes inevitable. We carry it upstream to the start of a life, the mother and the child, and outward to the home, until healthcare is no longer a place you are taken but a system that comes to you. Build it or fund it, it is the same bet: that human flourishing, made economically inevitable, is the largest opportunity in medicine.
Put as a sequence: it starts with the one aligned unit, the clinic, and installs the recursive learning loop. That loop gives patients agency, providers visibility, and payers accountability, and it compresses cost, so the unit economics close and the right care becomes the cheapest care. From there the unit replicates, moves upstream, and distributes into the home, until the centralized hospital is displaced by distributed prevention. The result is human flourishing, reached by economics, not idealism.
- 01
The inversion
A hospital lives on sickness; the clinic lives on health, the one aligned unit.
- 02
A recursive loop
At its core, getting smarter with every patient it sees.
- 03
It compounds
Patient agency, a 360° view, payer accountability, and lower cost.
- 04
Economics close
The right care becomes the cheapest care.
- 05
It carries outward
Replicate the unit, go upstream to mother & child, and into the home.
- 06
Human flourishing
Distributed prevention displaces the hospital, by economics not idealism.
Healthcare that runs itself, starting where it has to work first.
Team
The people building it.
We are a talent-dense, high-agency group of engineers and researchers from the top IITs, with experience across Microsoft, Oracle, Amazon, IBM, Goldman Sachs, Fractal, and BharatGen, research at Harvard, MIT, and CAISH / Redwood Research, and publications at NeurIPS, ICLR, ACL, and EMNLP.
01Leadership
ASDr. Ashu SyalPhysician and Medical Director
Pediatrician and the clinical anchor of our practices; Medical Director at Nag Clinics and Clinical Assistant Professor at Baylor College of Medicine and UTMB, with training in pediatric hematology and nephrology. The physician who chose independence.
Pediatric NephrologyBaylor & UTMBMedical Director
Rahul BhardwajChief Financial Officer
IIM Calcutta alumnus leading finance, acquisitions, clinic economics, and capital planning for the operator-plus-software model.
IIM CalcuttaM&Aclinic economics
Pratham SinghChief of Staff
IIT Bombay operator coordinating CEO priorities, clinic operations, hiring, and product execution across the team.
IIT BombayopsCEO office
02Models and research
Ankan BiswasHead of Models
IIT Kharagpur '23; ICLR/NeurIPS researcher; ex AI Research at Fractal and MIT; leads fine-tuning, distillation, and agentic frameworks.
IIT KGPICLR/NeurIPSex Fractal/MIT
Rishabh KumarModels and voice
AI researcher specializing in ASR, speech-LLMs, and multi-agent systems, with papers at ACL, EMNLP, EACL, and Interspeech. PhD at IIT Bombay, B.Tech from IIT Dhanbad, and BharatGen experience, working on our models and the patient-agent stack behind Bishop.
PhD AI, IITBASR · Speech-LLMACL/EMNLP/Interspeech
Tasmay Pankaj TibrewalResearch and models
IIT Kharagpur AI researcher and incoming MARS Fellow with CAISH / Redwood Research; past work spans Fractal and Harvard.
IIT KGPRedwood/CAISHFractal/Harvard
03Agents
Kirti BagriHead of Agents
Founding AI engineer and IIT Delhi CSE grad; Oracle SDE-2 / Analytics background; owns RCM from claims to denials and appeals.
IIT Delhi CSEOracle SDE-2Founding AI
Rashmitha PittalaScribe
IIT Delhi CS '23 engineer building the AI that converts physician-patient conversations into structured notes, summaries, and billing codes.
IITD CS '23Scribe leadclinical docs
Rushi ChavdaGuardian
IIT Bombay AI researcher with Harvard Research / IBM, EY, Reliance Jio, and Birlasoft experience; tracks care, adherence, and quality measures over time.
IIT BombayHarvard/IBMEY/Jio/Birlasoft
CPChappidi PreethiCredentialing
IIT Bombay CSE '26; AI intern at SproutsAI, ex-CV intern at VISIST.AI, and AI Engineer at THP; works across LangChain, RAG, PyTorch, and computer vision.
IITB CSE '26SproutsAIex VISIST.AI
Udita MittalRCM
IIT Delhi data scientist with Microsoft experience; previously built Castle, our practice-acquisition marketplace, and now works on RCM alongside Kirti, from claims to denials and appeals.
MicrosoftIIT DelhiRCM
Simran BhagatRCM
IIT Delhi engineer with Goldman Sachs and fintech-payments experience; works on RCM alongside Kirti, from claims to denials and appeals.
Goldman SachsIIT DelhiRCM
Yash ChoudharySupply chain
IIT Bombay economics undergrad building the Supply chain agent: clinic inventory, ordering, and demand prediction across the operating network.
IIT BombayEconomicsSupply chain
04Data and infrastructure
Ajay EdupugantiHead of Data
IIT Madras builder with Lincecraft AI background; leads Foundry's shared data layer, knowledge graphs, ontologies, and pipelines.
IIT MadrasLincecraft AIFoundry
Nadeesh GargFoundry
Prophecy data engineer and IIT Delhi alum building scalable data products and infrastructure for the shared record under every agent.
IIT DelhiProphecydata infra
Charan KumarFoundry
Amazon SDE II, ex-Flipkart, and IIT Bhubaneswar CS; builds backend and data infrastructure that keeps the shared record reliable.
Amazon SDE IIex FlipkartIIT Bhubaneswar
05Join us
We are hiring across research, engineering, and operations, in Houston and Bangalore. If building the AI that runs a clinic, and carrying good healthcare to far more people, is the kind of problem you want to spend your time on, we would love to hear from you at admin@independenceos.ai.