September 1, 2026
AI Automation for ABA Practices: Where to Actually Start
By Andrew Hendrich, a former Senior RBT who consulted BCBA leadership on clinical operations, now building Anomaly.
Search "AI for healthcare" and you'll get content written for hospital systems with a six-person IT department and a six-figure software budget. That's not useful if you're running a 20-person ABA practice trying to keep RBT schedules staffed and certifications current. The advice doesn't transfer, and most of it isn't even trying to.
Here's what we actually see when we sit down with an ABA practice and map where the hours are going.
Start with admin, not clinical
The fastest, lowest-risk automation wins in an ABA practice are almost never on the clinical side, and that's by design, not limitation. Client treatment data is PHI, which means any automation touching it needs a signed Business Associate Agreement, tighter security review, and a much longer runway before it's live. Most practices don't need to start there to see real time back.
The bigger, faster win is almost always administrative: RBT hiring and onboarding, scheduling and shift coverage, supply ordering, and increasingly, credentialing and compliance tracking. That's the same order we work through on every audit.
RBT credentialing is the quiet problem getting worse
New RBT certification standards introduced a two-year renewal cycle with accumulating PDU/CEU requirements, a minimum 5% supervision-time documentation rule, and annual competency assessments. Unlike BCBA credentialing, there's no automated portal for RBTs, so most small agencies are tracking certification numbers, expiration dates, supervision logs, and CEU proof in a spreadsheet, manually, per technician.
That's pure staff-side data (zero client PHI), which makes it one of the safest, most overlooked automation targets in the entire practice. And the stakes are real: a lapsed certification doesn't just create paperwork, it can pull a technician off a client's care until it's resolved. Here's a real credentialing system we built for exactly this problem.
Where not to start
Billing, claims, and prior authorization get recommended constantly as "easy AI wins" in generic healthcare content. For an ABA practice, that's usually the wrong first move. That data is tied to a named client's treatment, which means it's PHI, which means a BAA and a much heavier compliance lift before you automate anything. Not impossible, just not where the fast, safe ROI is.
The actual first step
Before automating anything, map what's actually eating the hours, not what a generic checklist says should be automated. That's the entire point of an audit call: no assumptions, just what your practice specifically is losing time to, prioritized by what it'll actually save you.
If you're running an ABA practice and RBT hiring, scheduling, or credentialing tracking is eating more hours than it should, tell us about your practice and we'll show you exactly where.