ASA Classification
Disposition & Required Workup
Medication Holds
Anesthetic Considerations
engine vX.X · clinical-config vX.X
reviewed and attested by [Physician], MD · [date]
reviewed and attested by [Physician], MD · [date]
PeriGate does two jobs in one two-gate workflow. It runs a deterministic clinical engine across pre-anesthesia evaluation — risk scores, medication safety, guideline timing — so your PAT clinic can clear, work up, or defer with confidence. Then it generates the chart-ready pre-operative note, assembled from sections a physician has individually reviewed and attested.
Same-day cancellations and preventable delays are among the most expensive failures in surgical operations — an empty OR slot, a wasted anesthesia team, a patient sent home. Most trace back to the same root cause: pre-anesthesia evaluation that is manual, inconsistent, and disconnected from current guidelines.
For ambulatory surgery centers the math is sharper still: a same-day cancellation is unrecoverable revenue on a schedule with no slack.
PeriGate is built on a principle most clinical AI products avoid: the physician is not a reviewer of last resort — the physician is the gate. And the gate is engineered to be light: the system does the reading, the math, and the writing; the clinical team does only the judgment. Review by exception, not click-through-everything.
Patient context arrives via FHIR from your EHR. Problems, medications, labs, and the surgical booking — normalized into a structured pre-anesthesia case. No chart hunting.
A deterministic clinical engine — not a language model — evaluates risk scores, medication safety, and guideline-based timing. Rules are versioned, auditable, and never improvised. AI drafts the judgment sections: the narrative reasoning a physician would otherwise type. AI is never permitted to override the engine.
A nurse or clinician confirms the case as assembled — the system has already done the chart-pulling. Garbage in stops here.
Review by exception: clean sections clear in a single action, while flagged sections — safety-critical findings, low-confidence drafts — demand individual attention and can never be bulk-accepted. Every decision is recorded.
A chart-ready pre-operative assessment note, assembled deterministically from the attested sections, with engine values reproduced verbatim and a provenance block recording exactly what was signed, by whom, under which engine and configuration version. Exports as text or FHIR DocumentReference.
Your EHR already organizes perioperative care as pre-, intra-, and post-operative. So does the anesthesia record, the quality committee, and the malpractice review. PeriGate is built to that same spine — starting where the decisions are made furthest in advance, and where a wrong call is still recoverable.
Pre-anesthesia evaluation and testing: the clearance decision, the required workup, medication holds, and the signed pre-operative note.
The anesthetic plan and the record that documents it. The same architecture applies: the engine computes, the anesthesiologist attests, the record carries the lineage.
Disposition and the handoff. Where the patient recovers is a decision made before incision — and it belongs in the same signed record.
Status is stated plainly and on purpose. Phase 01 is the product. Phases 02 and 03 are declared roadmap — described here so you can evaluate the architecture, not so you can buy something that does not exist yet. We will tell you the same thing on a call.
A growing class of tools will summarize a chart and hand a physician a document to sign. A summary, however fluent, answers to no one — and a signature on the whole document records nothing about which parts the clinician actually agreed with.
PeriGate is built the other way around. Nothing reaches the chart that a physician has not individually signed — section by section, not document by document. The deterministic engine provides the ground truth; the AI drafts only judgment narrative; the physician accepts, modifies, or overrides each section; and the final note carries a complete provenance record — engine version, clinical-configuration version, and the per-section lineage.
The result is a document your compliance office, your quality committee, and your malpractice carrier can actually interrogate — six months later, on a case none of you remember.
Risk scores and safety logic come from versioned rules, never from a language model's recollection.
The model never learns from live clinical data. Improvements ship as discrete, re-validated version releases.
Every section of every note records who reviewed it and what they decided — accept, modify, or override.
Including us. This category is filling up with products that describe themselves in nearly identical language, and the differences that matter are not visible from a homepage. These are the questions we would ask.
A signature on a whole document is not the same as a signature on each finding. Ask whether the clinician can accept part of the output and reject the rest — and whether that decision is recorded anywhere. PeriGate attests per section, and flagged sections can never be bulk-accepted.
Your quality committee will ask. The answer requires a version-pinned engine, a version-pinned configuration, and a record of what the reviewer changed. PeriGate writes that block onto every note.
Guidelines are revised. Ask which societies, which editions, and how long it takes a new publication to reach the product. PeriGate treats guideline currency as infrastructure: a quarterly currency check and an immediate update on new society guidance, both versioned.
Performance claimed after the fact is not performance. Ask whether agreement thresholds were defined before measurement, whether they are reported per feature or only in aggregate, and whether safety-critical findings have their own floor. Ours were locked before we measured, and we will show you the thresholds.
A product that claims every phase, every risk factor, and every integration is describing a roadmap in the present tense. Ask for the boundary. Ours is on this page: pre-operative today; intra-operative and post-operative are declared roadmap; no live EHR deployment yet.
Integrates. Doesn't add a program.
PeriGate is designed for SMART-on-FHIR launch inside the EHR workflow your anesthesia and PAT teams already use, and mapped to the perioperative phases your chart already models. No parallel portal, no swivel-chair workflow. Deployment follows the standard Epic vendor pathway.
Built for the schedule with no slack.
Ambulatory surgery centers feel every cancellation. PeriGate gives anesthesia groups a consistent, guideline-current clearance workflow with a lightweight FHIR ingestion path — designed for the realities of ASC IT, not just academic medical centers.
PeriGate is architected as clinician-directed clinical decision support: the physician independently reviews the basis for every recommendation and attests every output before it is used. This two-gate design is aligned with the clinical decision support criteria described in Section 3060 of the 21st Century Cures Act. The system is conservative by default — where uncertainty exists, it recommends more workup and more monitoring, never less.
We are building the evidence base for attested clinical decision support the way it should be built: prospectively defined thresholds, per-feature concordance measurement, and clinician attestation on every case. We partner with anesthesia departments, private anesthesia groups, and ASCs on retrospective, de-identified validation studies — IRB-governed, with co-authorship on resulting publications.
If your group runs a pre-anesthesia clinic and wants a hand in shaping this category, we want to talk.
PeriOp AI was founded by Michael Grinn, MD MPH, a practicing cardiac anesthesiologist, with technical co-founder Farhan Baluch. The clinical advisory board is in formation.
Demos are live, guided walkthroughs of the product — no patient data, no marketing pitch. We'll follow up within two business days to schedule.