"Triage" gets used loosely for any phone-answering. It shouldn't be. Assessing symptoms and deciding disposition over the phone is clinical practice, and who may do it is a scope-of-practice question with liability attached. Here's the line, per the people who insure practices, and where an AI vendor honestly fits.
The line
The workable middle ground for unlicensed staff, per carrier guidance: information collection under yes/no algorithms with a licensed provider always reachable for anything judgment-shaped. Collection is not assessment; the moment someone weighs a symptom, they're triaging.
We build AI phone answering for medical practices, so read this knowing that. Our design rule is absolute: Hello AI never assesses a symptom, never reassures, never advises. It opens after-hours clinical calls with a 911 deflection, asks the caller whether their matter is urgent or routine (the caller decides, not the model), and then does the thing software IS good at: reaching your on-call clinician relentlessly, with an accepting cascade, backups, an SMS fail-safe, and a timestamped log of every attempt. If an AI vendor tells you their bot "triages" patients, ask them who holds the license.
FAQ
No. Triage requires interpreting symptoms and making independent clinical judgments, which sits outside the medical assistant scope of practice; risk-management guidance from malpractice carriers is explicit that unlicensed personnel should not triage by phone or give telephone advice. MAs can collect information under strict yes/no protocols with immediate escalation paths to a licensed provider, which is a different activity from triage.
Operators are unlicensed, so no, and reputable services don't claim to. What they legitimately do is message-taking and relay under rules you define. The risk zone is improvisation: an operator who reassures a worried caller or downplays a symptom has just performed unlicensed triage on your behalf, with your liability attached.
Our position, as an AI answering vendor: no, and vendors who imply otherwise are creating risk for their customers. Hello AI is deliberately designed NOT to triage. It deflects emergencies to 911, asks the CALLER to choose urgent or routine, and routes accordingly: urgent to your on-call clinician, routine to the morning queue. Assessment stays with licensed humans; the AI only moves the call to them faster.
Emergency disclaimer first (dial 911), then structured information gathering without assessment, then routing by explicit rules: urgent matters reach a licensed clinician (your on-call provider) promptly and verifiably, routine matters queue for business hours, and every step is documented with timestamps. The documented part matters: in a dispute, 'we paged the doctor' is a claim; a timestamped escalation log is evidence.
This page is risk-management education, not legal advice; scope of practice varies by state. Sources: Coverys on MA scope of practice, MedPro Group on telephone triage policies, The Doctors Company on triage patient-safety strategies, Medical Mutual of Maine on triage systems.