The intelligence scribe for psychiatry
A scribe that knows psychiatry.
Sigmund sits in on the session and drafts the note — across the full range of mental health care: depression, anxiety, attention, mood, neurocognitive, and trauma. His clinical knowledge nodes hold a live, evolving understanding of the scholarly clinical literature and feed it into every note, so the documentation is sharper and the evidence is cited as it's written.
Free for clinicians. Free forever for trainees. See what's included →
Built with
In collaboration with
The gap
12.9%
of U.S. children with ADHD receive any medication. The AAP, AACAP, and FDA publish concordant treatment guidelines, yet 60–70% of pediatric ADHD cases in community settings are managed off-guideline. The evidence exists. The delivery system to the right decision at the right moment does not.
Sigmund distinguishes a methylphenidate trial in adults from one in preschoolers, by design — across population, intervention, outcome, effect size, and timing simultaneously. That distinction is the product.
A glimpse of Sigmund at work
An auto-playing demonstration of how Sigmund reasons about a real patient case. The voice is character-driven; the clinical content is real.
How Sigmund works for you
Sigmund sits in on the full clinical session — listening to the conversation, watching the mental status exam, weighing the risk, and keeping the evidence trail behind every decision.
Every recommendation arrives with the PMIDs. Click any line. See the studies behind it.
The signature line stays empty. The MSE stays editable. The clinician closes the loop.
Freed users report 30 minutes saved per clinic day. Sigmund matches that — with the citations.
Validation
Sigmund's clinical knowledge runs on a network of knowledge nodes that hold a dynamic, continuously updated understanding of the scholarly clinical literature across mental health — depression, anxiety, attention, mood, neurocognitive, and trauma. That understanding is integrated directly into the scribe, so every note is grounded in current evidence and cited as it is written. Generation 1 has been operating at Integrative Psychiatry Manhattan, a high-volume New York City outpatient practice, turning patient data, clinical guidelines, peer-reviewed evidence, and expert-committee judgment into citation-traceable, guideline-concordant recommendations.
87.4%
Diagnostic accuracy across five condition categories
100%
Clinical staff endorsement of improved work quality
400%
Increase in documentation captured per encounter
0
Patient safety complaints across the deployment
0.81–0.88
Effect size (r) versus generic AI baselines
R21 + R01
Two NIH grants behind CEBA — R21 (PAR-25-310) and R01 (PAR-25-283) — for prospective validation. Primary outcome: Guideline Concordance Score.
N = 124 charts · Integrative Psychiatry Manhattan · NIH R21 prospective validation underway.
In the note
An 8-year-old boy. ADHD with anxiety comorbidity. No prior treatment. As Sigmund drafts the note, it ranks the treatment options the clinician will document — by what the evidence supports:
Eight-year-old male presents with a nine-month history of inattention and academic decline. Mother reports difficulty completing schoolwork, frequent fidgeting, and new-onset evening anxiety with episodic refusal to attend school. No prior psychiatric treatment. No family history of stimulant use or substance use disorder.¹
ADHD, combined presentation (F90.2) with comorbid generalized anxiety disorder (F41.1). Treatment-naive. Acuity: routine.
The note arrives as a draft. The MSE, the differential, the citation chain — all surfaced by Sigmund. The clinician edits, signs, and finalizes.
How it works
Sigmund splits each paper into five section-level vectors — population, intervention, outcomes, effect size, and timing — and ranks treatments against the patient profile across every dimension simultaneously. Citation-traceable on every line.
Combined with evidence-tier weighting (RCT > meta-analysis > cohort), recency boosts, and effect-size aggregation, the result is a citation-traceable treatment ranking. No black box.
Traceable, always
Click any recommendation. The PMIDs unfold. Study design, sample size, effect size, year. The full evidence trail behind every ranking.
Atomoxetine · score 0.935 · rank #2 for this profile
Where the work happens
Sigmund works where you already do — voice-activated, chart-embedded, listening in the background. He learns your pace, your vocabulary, your clinical patterns.
The work psychiatrists actually do
The visit is the small part. Around it sits the documentation a psychiatrist actually owns: polypharmacy safety, controlled-substance compliance, measurement-based-care scoring, suicide-risk documentation, prior-auth fights, school and disability letters. Sigmund is built around the whole job, drawing on the deep psychopharmacology and child-psychiatry research base of the Sultan Lab.
Most psychiatric patients are on three to six medications. Sigmund surfaces drug–drug interactions, QTc-prolongation stacks, serotonin-syndrome risk, and CYP2D6 / 2C19 / 3A4 metabolizer-relevant pairings as the regimen is built — not after the prescription is sent.
Schedule II stimulants, benzodiazepines, buprenorphine, esketamine — each carries a documentation tail. Sigmund tracks PDMP checks, REMS-program lookups (clozapine, Spravato, isotretinoin), and DEA telemedicine-rule changes by state, and inserts the required language into the note automatically.
PHQ-9, GAD-7, ASRS, Vanderbilt, MDQ, AUDIT-C, C-SSRS, SCARED, PSC-17, CY-BOCS — captured at the appropriate cadence, scored as the patient speaks, trended longitudinally, and pulled into the note without a separate questionnaire workflow.
C-SSRS scoring, dynamic risk-factor capture, written safety plan, escalation-pathway documentation. Built to the standard a malpractice reviewer expects to see, generated from the conversation as it happens.
Atypical-antipsychotic step therapy. Brand-name stimulant appeals. Spravato auths. Lurasidone and cariprazine in pediatrics. Each letter drafted from the chart, with the supporting citation chain attached — ready to fax or submit.
504 plans. IEP requests. College accommodation letters. FMLA paperwork. Court-mandated-treatment compliance. SSI / SSDI documentation. The non-billable documentation that eats a psychiatrist's evenings, drafted from the chart in seconds.
Most psychiatric medications are not FDA-labeled for children. Sigmund flags off-label use, surfaces weight-based dosing, references AACAP practice parameters by age band, and documents the rationale the moment the decision is made.
When CYP2D6, CYP2C19, or CYP3A4 results live in the chart, Sigmund surfaces the implications — SSRI selection, antipsychotic dose adjustment, atomoxetine metabolism — in the prescribing moment, citation-chained to the evidence behind each call.
Each capability is grounded in the published psychopharmacology and child-psychiatry literature curated by the Sultan Lab — the same evidence base that informed the FDA's 2015 clozapine REMS modification and the 2025 JAMA Psychiatry analysis of ADHD-medication protective effects.
Psychiatry is a longitudinal specialty
A single note is a snapshot. Psychiatric care lives in the arc between notes: how this dose responded, what last winter's mood looked like, which medication trial failed and why. Sigmund builds the longitudinal record automatically — and surfaces the relevant history at the moment of the next decision.
The five dimensions psychiatrists hold in their heads between visits — surfaced as the chart opens. This is what clinical software built for psychiatry looks like.
In your day
Concrete daily wins. No more pajamas-and-charting at midnight.
Sigmund drafts as you talk. The note arrives in your inbox when the door closes behind the patient — not at 11 PM.
A complete mental status exam in your clinical voice, drafted from the conversation, ready to sign.
Every clinical claim arrives with the citation that justifies it. Notes that go to payers read like the textbook.
Sigmund listens while you stay present with the person across from you. The chart populates in the background.
Backed by NIH-funded research at the Sultan Lab · Columbia · NYSPI.
The roadmap
Sigmund's multi-vector architecture is condition-agnostic. ADHD is the index condition because the longitudinal evidence infrastructure exists today. The rest follow the same architecture.
Deployed at Integrative Psychiatry Manhattan. Validated at 87.4% diagnostic accuracy. R21 (PAR-25-310) prospective validation pending.
Graph-based RAG over the full ADHD clinical ontology. 18-month stepped-wedge implementation pilot across 12 community prescribers — including the nurse-practitioner cohort driving the post-2020 stimulant surge. Adds structured deprescribing support and family-engagement workflows.
Next engine. Largest pediatric and adult literature base. Natural extension of multi-vector architecture.
Builds on the population-derived risk framework. ASD-ADHD, anxiety-depression, substance use. The architecture was built for this from day one.
How Sigmund scales
Sigmund grows outward in phases. At every one, patient data stays on the home clinic's own local infrastructure. The system scales out; the data never leaves home.
Independent and group mental-health clinics. Running today at Integrative Psychiatry Manhattan.
Academic medical centers and their affiliated outpatient networks.
State-level clinical informatics and behavioral-health systems.
National health systems in the United States — and internationally.
However far Sigmund scales — from a solo practice to a national system — the model never changes. Sigmund runs inside each institution's own HIPAA-compliant infrastructure; identifiable data stays put, and on the rare occasion anything moves, it is de-identified first. Sigmund comes to the data; the data does not come to Sigmund.
About
Assistant Professor of Clinical Psychiatry, Columbia University Irving Medical Center · Attending Psychiatrist, NewYork-Presbyterian · Director, Sultan Lab for Mental Health Informatics · Medical Director, Integrative Psychiatry of Manhattan · Member, Columbia Data Science Institute · FDA Consultant.
Dr. Sultan is double board-certified in Adult and Child & Adolescent Psychiatry. He completed an NIMH T32 postdoctoral fellowship in pharmacoepidemiology under Mark Olfson, MD, MPH, and is currently completing the NIDA K12 Mentored Clinical Scientist Development Program in clinical natural-language processing and machine learning under Frances Levin, MD (Columbia) and Timothy Wilens, MD (Harvard / Massachusetts General Hospital).
His published work has shaped national psychiatric practice on three fronts. His 2017 analysis in the Journal of Clinical Psychiatry provided the empirical foundation for the U.S. modification of the clozapine REMS, expanding access to a uniquely effective treatment for treatment-resistant schizophrenia. In 2026 he was first author on an invited editorial in JAMA Psychiatry — "Protective Effects of ADHD Medication — Diminishing Returns as Treatment Broadens" — that opened a new evaluation paradigm by moving the field from symptom-rating-scale change to lived real-world outcomes. His 2026 JAMA Psychiatry paper on cannabis exposure and new-onset psychosis in young people, his 2023 JAMA Network Open paper on nondisordered cannabis use (top 1% of all articles by Altmetric attention), and his 2019 JAMA Network Open paper on antipsychotic prescribing in youth with ADHD (taught in residency programs nationally) anchor a sustained portfolio in psychopharmacology epidemiology.
His work has reached the public-health and policy conversation at scale, with citations or quoted commentary in more than 100 national and international outlets — including The New York Times, NBC News, NPR, Time, Scientific American, Fox News, U.S. News & World Report, and Forbes Health. He serves on the AACAP Pharmacology & Neurotherapeutics Committee (2023–present), as a Reviewing Editor for BMC Psychiatry (Springer Nature), as an NIH Study Section reviewer, and as a peer reviewer for the American Journal of Psychiatry, British Journal of Psychiatry, JAMA Network Open, JAACAP Open, and the BMJ Group. He has been an invited plenary speaker at the American Professional Society of ADHD and Related Disorders (APSARD, January 2026; "Do Stimulant Medications Prevent Substance Use?") and at the Columbia Data Science Institute (March 2026; "AI Companions and the Future of Care").
Dr. Sultan directs the Sultan Lab for Mental Health Informatics — seven researchers at Columbia — and leads a ten-clinician outpatient practice that doubles as a translational deployment site with a longitudinal database of more than one thousand patients. Sultan Lab mentees have matriculated to U.S. allopathic medical schools at Icahn School of Medicine at Mount Sinai, Drexel University College of Medicine, and Weill Cornell Medicine, and are co-authors on Sultan Lab publications in JAMA Network Open, JAMA Psychiatry, Pediatrics, and the American Journal of Preventive Medicine.
Sigmund is built from a deeply held conviction: the gap between what the evidence supports and what reaches the patient is a solvable problem.
Built-in guardrails
Every one of these is a property of how Sigmund is built, not a policy you have to trust us to follow.
Your data is never used to train or improve our models.
Sigmund runs inside your institution's existing HIPAA-compliant environment. Identifiable patient data never has to leave it.
Nothing reaches the chart until you review and sign.
Adjunctive support. The decision is always yours.
Every output cites its guideline and source.
All processing runs under signed BAAs.
Engineered for safety
Everything Sigmund touches is HIPAA-compliant — but compliance is the floor, not the story. Sigmund runs inside your institution's own environment, where the data already lives, and minimizes how much data moves at all. Identifiable PHI never has to cross into a new system — a stronger privacy posture than any scribe that ships your patients' data out to a vendor cloud.
Sigmund deploys inside your own HIPAA-compliant infrastructure — running today inside Integrative Psychiatry Manhattan, and built for academic systems like Columbia and NewYork-Presbyterian. By default nothing leaves; an institution can optionally allow de-identified-only calls to US frontier models, under a no-retention, no-training contract.
AES-256 at rest. TLS 1.3 in transit. Audit-logged on every access. No exceptions.
Sigmund is investigational and intended to assist — not replace — clinical judgment. Patient data handling reviewed by the Sultan Lab for Mental Health Informatics at Columbia University Irving Medical Center.
Common questions
What does it cost?
For most clinicians, nothing. The core tools — initial patient analysis, symptom tracking, and the AI scribe — are free, and free forever for verified trainees. We don't run ads and we don't sell your data; the free tier is funded by grant-backed research infrastructure and by a paid Practice tier (the full decision-support engine) launching Q3 2026. Details on the pricing page.
Will it work with the systems I already use?
Yes. Sigmund connects to the clinical informatics infrastructure your practice already runs — including Epic, athenahealth, Practice Fusion, Valant, AdvancedMD, ICANotes, and Tebra — reading the chart and writing the note back into it. We've begun connecting these at our design-partner clinic. If the system you use isn't listed, tell us in the waitlist form and we'll prioritize it.
What if I leave Sigmund?
Your notes are yours. Export to PDF or structured JSON at any time, in bulk. We don't lock data; we keep it portable by design.
Does it handle telehealth visits?
Yes. Sigmund captures the visit whether the patient is across the room or across the country. Same MSE precision, same citation chain, same auditability.
Who is responsible for the clinical content of the note?
You. Sigmund drafts; the clinician signs. The signature line stays empty until you complete it. Sigmund is clinical decision support, with the clinician in authority on every recommendation.
Read further
Long-form references for the architecture, the documentation work, the decision moments, and the wider AI-scribe field.
Beyond the scribe
Appropriate ADHD treatment, behavioral activation, CBT, clozapine compliance — and real-time monitoring through PAWS.
Pillar
What it requires, how CEBA implements it, and the NIH R21 validation study.
Documentation
MSE, progress notes, medication management, risk assessment, controlled substances.
Decisions
ADHD with anxiety, stimulant titration, antidepressant selection, benzo deprescribing, pediatric prescribing.
Compare
General-medical scribes, psychiatry-specific scribes, and where Sigmund fits.
Resources
Honest answers, transparent pricing posture, and a free clinical resource you can use today — no signup gate.
FAQ
Peer-review status, HIPAA posture, clinical-systems integration, competitor comparison. Direct answers.
Free download
Full MSE template — fill-in form, printable, no signup. The template our clinic actually uses.
Pricing
Honest about what we can tell you now, what we are holding back, and what we will never charge for.
Founder
Built by Dr. Ryan Sultan — Columbia psychiatrist, Sultan Lab director, founder of Integrative Psychiatry Manhattan.
Sigmund's core tools are free for mental health clinicians and free forever for trainees. We onboard by hand so we can sign a BAA and set you up properly. Tell us about your practice or training program — we'll be in touch within a week.
We'll be in touch. In the meantime, take a look at how it works.
Connects with the systems you already use
Epic·athenahealth·Practice Fusion·Valant·AdvancedMD·ICANotes·Tebra