Clinical AI grounded in facts — not hallucinations.
Most AI scribes write notes straight from a transcript and hope the model doesn't improvise. Notat's patent-pending FactsContext™ engine understands the encounter first: it extracts structured clinical facts, generates documentation from those facts, and shows you the raw facts to do with as you please. It works across real multilingual care: spoken capture in 99+ languages and a product translated in 15 languages.
Understands before it writes
FactsContext™ separates understanding from writing. That separation is what makes the documentation trustworthy — and what makes every downstream capability possible.
1. Conversation
The encounter happens naturally. Notat listens ambiently — no dictation, no prompts, no screens between you and the patient.
2. Clinical facts
FactsContext™ extracts structured clinical facts from the conversation: symptoms, findings, medications, decisions, plans. Each fact is anchored to what was actually said.
3. Documentation
The note is generated from the extracted facts — never straight from a raw transcript. If a statement isn’t supported by a fact, it doesn’t belong in the note.
4. Review
You see the raw facts next to the note. Verify any sentence against its source, edit what you want, and sign with confidence. The clinician always remains in control.
Radical transparency: you see the raw facts
Trust isn't a promise — it's something you can check. Notat exposes the structured medical context behind every note, so "where did this sentence come from?" always has an answer.
The facts are yours
Notat shows you the raw extracted medical facts from every encounter — not just the polished note. Inspect them, verify the note against them, or reuse them however you please.
Every sentence is traceable
Because documentation is generated from structured facts, every statement in the note traces back to something that actually happened in the encounter.
Nothing invented
Transcript-direct AI fills gaps with plausible-sounding text. FactsContext™ is designed to reduce unsupported statements by only writing from verified clinical facts.
Everything built on one clinical understanding
Because FactsContext™ produces structured facts — not just a note — the same understanding of the encounter powers every capability in the platform.
Clinical notes
SOAP, H&P, progress notes — written from facts in your style.
ICD-10 coding
Code recommendations with the supporting evidence excerpt attached.
Referral letters
Generated from the same clinical understanding, not re-dictated.
Magic Edit
Rewrite, translate, or restructure — the facts stay the source of truth.
Patient instructions
Plain-language follow-up built from what was actually decided.
Evidence answers
Clinical questions answered in the context of the encounter facts.
Validate the architecture
FactsContext™ connects the proof methodology, vendor comparisons, ICD-10 evidence, and multilingual workflows back to the same facts-first architecture.
Clinical AI evaluation
See the transparent methodology Notat uses to evaluate unsupported statements, fact coverage, and correction burden.
Open
Compare AI scribes
Review dated vendor comparisons across hallucination safeguards, languages, compliance, and EHR fit.
Open
ICD-10 evidence
Explore condition guides and code suggestions that attach evidence excerpts to recommended codes.
Open
Multilingual scribe
See 99+ spoken-language capture and translated workflows for cross-language care.
Open
Frequently asked questions
What is AI hallucination in healthcare documentation?
Hallucination is when an AI writes plausible-sounding statements that were never said or observed — an invented medication dose, a symptom the patient never reported, a plan that was never discussed. In clinical documentation, unsupported statements are a patient-safety and legal problem, not just a quality problem.
How does FactsContext™ reduce hallucinations?
Most AI scribes generate notes directly from a transcript, which lets the language model improvise. Notat’s patent-pending FactsContext™ engine works in two distinct stages: it first extracts structured clinical facts from the encounter, then generates documentation only from those facts. Statements without a supporting fact are designed not to appear in the note.
Can I see the facts behind my note?
Yes. Notat shows you the raw extracted medical facts alongside the generated note. You can verify any sentence against its source facts before signing, and reuse the facts for coding, referrals, or anything else.
Does the clinician still review the note?
Always. FactsContext™ makes review faster and safer because you can check the note against structured facts instead of re-listening or guessing — but you review and sign every note. The AI does the writing; you do the medicine.
Does FactsContext™ work in my language?
Yes. Notat supports spoken capture in 99+ languages through its AssemblyAI-powered speech layer, and the product is translated in 15 languages: English, Norwegian, Danish, Swedish, Finnish, Estonian, Dutch, German, Spanish, Italian, French, Arabic, Polish, Portuguese, and Hindi. The conversation, the extracted facts, and the final documentation can each be in different languages, including cross-language visits where the patient and the record don’t share a language.
See the facts behind your next note.
Record a consultation, watch FactsContext™ extract the clinical facts, and read a note you can verify line by line. You review and sign every note — Notat just makes that trivially easy.
Try Notat free