Essays on clinician-governed AI
Notes, Intake, Triage: What a Provider API Does All Day
The real AI product in healthcare is not a chatbot. It is the hours it gives back.
Forget the chatbot demo. The product that matters in healthcare AI is infrastructure: the unglamorous hours that eat a clinician’s day and never appear in anyone’s keynote. Documentation. Intake. Triage. The in-basket with forty messages and one emergency hiding inside it. This is where a provider API earns its keep, and it is what Antiphon is being built to do.
Start with clinical notes. Documentation is the shadow workload of modern practice, hours of writing about the work instead of doing it. A supervised API drafts the note from the session itself, structured the way auditors and insurers expect, and the clinician reviews, corrects, and signs. The clinician remains the author. The machine is the scribe.
Move to intake and assessment. Every new patient should get the same careful, complete intake, but in a busy practice the intake is whatever survives the schedule. An API administers validated instruments, PHQ-9, GAD-7 and their kin, the same way every time, scores them instantly, and hands the clinician a structured picture before the first hello. Consistency is a clinical virtue, and machines are consistent.
Then the working day itself. Diagnosis support: differential suggestions anchored to criteria, for the clinician to accept or overrule. Treatment planning: evidence-based plan drafts tied to the assessment, editable by the person who carries the license. Triage: in-basket prioritization that surfaces the urgent message in the pile of forty, and the morning pre-chart digest that tells a clinician, before the first appointment, exactly what today holds. Billing guidance: coding support so the practice is paid for the work it actually did.
None of this is a consumer toy. It is a HIPAA-compliant API, plug and play, that lets practitioners build purpose-specific applications for their own practices with no engineering required. And every one of these uses sits under the accountability architecture from our first essay: outputs labeled as AI-generated, clinical actions traceable to a reviewing clinician, audit trails a compliance officer can read.
The market here is not the lonely teenager at midnight, though he matters too. It is the 1M+ licensed mental health providers worldwide, drowning in administrative work while their waitlists grow. Sell them leverage, not companionship. The winners of “muse for healthcare” will be the teams that give clinicians their Tuesday mornings back.