Therapy software has been a tax on the work for two decades — built for the billing department, not for the session. SignalEHR exists because the people doing the most important conversations in the room deserve a tool that respects them. One platform. The clinical AI in the room, the documentation in eight minutes, and the front desk that runs itself.
Why SignalEHR exists
Three short paragraphs. We owe you the honest version.
The first thing
Therapists are drowning in administrative work. The average private practice clinician self-reports forty-seven minutes per session spent on documentation, scheduling, and billing — for a fifty-minute session. Most of that happens after the last client leaves. The result is a profession quietly burning out at one of the highest rates in healthcare, in a market where demand has never been higher.
The second thing
Legacy EHRs were built for billing departments at large hospital systems, then sold sideways to private practices that don't resemble those workflows in any meaningful way. The interface optimizes for the CFO, not the room. The features that get roadmap priority are the ones that maximize claims throughput, not the ones that help a clinician hear what their client just said. That bias is in every screen.
The third thing
AI finally enables what software promised in 2005 and never delivered. Real-time voice analysis that surfaces emotional signal without breaking the alliance. Clinical documentation that produces a finished, signed note in roughly eight minutes instead of forty-seven. A front-desk agent that handles the phone, the SMS, the calendar, and the insurance claims while you do the work nobody else can. That is what SignalEHR is. It is also, we think, the floor for what a modern EHR should be.
How we build
We say no to a lot of things. These are why.
Principle 01
The clinician is the authority on every clinical decision. SignalEHR surfaces patterns, drafts documentation, and clears the administrative path — but it never signs the note, never sets the diagnosis, never overrides the room. Every output is a draft for a human who knows the client. That is not a limitation of the AI; it is the design.
Principle 02
A platform that ships features nobody uses is a platform with nothing to say. We measure ourselves on the metrics that matter to a private practice — minutes saved per note, claims paid first-pass, no-shows prevented, slots filled from the waitlist. If a feature does not move one of those numbers, it does not ship. Quietly, ruthlessly.
Principle 03
No live-session audio storage. AES-256 in transit and at rest. SOC 2-aligned infrastructure. Signed BAAs. Audit logging on every PHI access, every time. We treat HIPAA, PIPEDA, PHIPA, Law 25, HIA, and SHIELD as the floor, not the ceiling — because the people sitting across from a therapist deserve nothing less.
The arc
We didn't start with a product. We started with the room, the calendar that didn't sync, and the note that wasn't finished at midnight.
Year zero
Hundreds of hours with licensed therapists in the US and Canada. The same complaint, in different words: I love my clients. I do not love my software. I do not love staying up until eleven writing notes.
Foundation
We refused to build another integrations marketplace. Calendar, billing, telehealth, intake, claims, AI documentation, and the front desk live in the same database — because that is the only way the workflows actually fit together.
Today
During a live session SignalEHR measures the voice itself (pitch, pace, energy). A language model reads those measurements alongside the transcript and shows affect shifts while the session is still running. Risk language gets raised to the clinician as decision support. The neuro-fuzzy layer we built for that job runs beside it. It does not drive what you see yet, because we are still measuring it. AI notes in seven formats finish in roughly eight minutes. Amelia handles the calendar, the SMS, the claims, and the phone. The point of all of it is a therapist who goes home on time.
By the numbers
Numbers we don't love to brag about. We're sharing them because the alternative — generic claims about “enterprise grade” — is exactly the kind of thing we're trying to replace.
1,000+
API endpoints
The full clinical, scheduling, billing, and front-desk surface across 900+ unique paths — one platform, one source of truth.
140+
frontend pages
Every workflow a private practice runs, from intake to insurance reconciliation, lives inside SignalEHR.
9,500+
automated tests
Unit, endpoint, and real-Postgres suites. The deploy build runs all three and stops if one fails.
HIPAA + PIPEDA
compliant
Plus PHIPA, Law 25, HIA, PIPA, CMIA, SHIELD, and SOC 2-aligned infrastructure underneath.
Who's building it
You are trusting us with clinical records. You should know exactly who we are, where we are, and how to reach us.
Sudipta Sarkar
Founder & CEO
Credentials
Sudipta built SignalEHR the way he spent seven years building infrastructure: assume it has to work when someone is depending on it.
Sudipta Sarkar is the founder of SignalEHR, a clinical-intelligence EHR for licensed therapists in the United States and Canada. It was built in Edmonton, Alberta, and runs on Google Cloud Run. The clinical scribe listens ambiently and does real multi-speaker diarization, so a group or couples session comes back attributed rather than blended into one undifferentiated voice. An AI receptionist answers chat, SMS and phone using the clinic's own operational knowledge. And the insurance billing crosses the border in one place, through Stedi in the US and TELUS eClaims in Canada.
He also built the clinical modelling layer underneath it: an adaptive neuro-fuzzy inference system — ANFIS — that tracks a session's emotional state as a continuous trajectory rather than a label per utterance. It sits behind a serving switch. Today the production default still serves the language model, and the neuro-fuzzy engine runs alongside it being measured. That switch stays where it is until the measurements say otherwise. It is a clinical decision, not a release note.
Before founding the company he spent seven years as an engineering manager at EPCOR, Edmonton's regulated utility, leading a multidisciplinary group of more than twenty engineers and technologists through sub-discipline leads, accountable for a multi-million-dollar annual programme. He was also a Responsible Member under the company's APEGA Permit to Practice, which put the professional conduct of that group's engineering work in his own name. Not the company's. His. Seventeen years of engineering sit behind that.
He then designed and built the entire platform himself: the clinical AI layer, the US and Canadian billing engines, the web console, and the native mobile apps. Bootstrapped, with no outside capital. He is also the company's Privacy Officer, and he signs every HIPAA Business Associate Agreement and Canadian Data Processing Agreement personally.
Regulated engineering has a habit that healthcare software mostly lacks: when something fails, a named person answers for it. That is the standard he builds to here. The AI drafts the paperwork and the clinician makes every decision, and the record shows which did which. Ask him where the model is weak. He keeps that list, and it is specific.
SignalEHR is a small, product-led team. Engineering, clinical research, and support all report to the founder — which is why the person who answers your support email is close enough to the code to actually fix the thing. We're hiring.
17 years
engineering & delivery leadership
20+ team
led in a regulated utility
1,000+
production API endpoints, built solo
Bootstrapped
no outside capital to date
Company of record
The promise
“Therapy is the most important conversation a person will have all week. The software around it should be invisible — present when it's useful, silent when it's not, and never the reason a clinician stays up until eleven.”
— Sudipta Sarkar, Founder & CEO, SignalEHR
Fourteen days. The full product. No credit card. A real human will help you migrate from whatever you're on now if you want it.