AI employees for ABA clinics
The leak grows at every ABA location you can't see into.
Authorizations expire unnoticed, denials stack up, and new families stall in intake. Relay builds custom AI employees that work on top of the EHR your ABA group already runs (CentralReach, Motivity, AlohaABA, RethinkBH, and the rest of your stack). They automate back office and front office workflows across every site, then hand each finalized action to a staff member to approve.
The AI employees are healthcare-specific, human-in-the-loop, and built to work on top of your stack without ripping anything out. At Sensory Speech and Occupational Therapy, a multi-location pediatric therapy group, the AI employees drove 100% claim accuracy and staff 33% more productive. In 30 minutes we'll show you exactly where it's leaking.
Where your ABA clinic is leaking revenue.
Seven places the back office leaks at every location. Open the ones that sound like your clinic.
Authorization management is the most-cited operational failure in ABA. It fails six predictable ways: hours that burn down unused, hours billed past the cap, authorizations that expire mid-treatment, reduced units at re-auth that nobody flags, reactive tracking instead of proactive, and no clear view of supervision and parent-training hours.
An AI employee keeps a live auth tracker for every active client, pulled from the EHR you already run. It flags units nearing expiration 30 days out and projects utilization against the current schedule so underused hours get caught before the re-auth window. It assembles the renewal packet (treatment plan, BCBA notes, prior auth number) for a billing coordinator or BCBA to review and submit.
A staff member approves every submission and every escalation. The AI removes the assembly and tracking burden; it never submits on its own. Initial ABA authorizations require a DSM-5 ASD diagnosis, a functional behavioral assessment, and a treatment plan with proposed hours and CPT codes (97151, 97153, 97155, 97156). Re-authorizations recur roughly every 30 to 90 days depending on payer.
Underused and expiring hours get flagged in time, so expired-auth write-offs stop accumulating.
ABA denial rates run structurally higher than broader healthcare. Authorization complexity, RBT-credential payer rules, and documentation standards create more failure surfaces. Each denial means a biller has to read the reason code, decide whether to correct or appeal, gather supporting materials, and resubmit inside a payer appeal window that ranges from 30 to 180 days.
An AI employee reads incoming denial EOBs and ERAs and classifies each one by type (payer-specific rule, documentation gap, auth issue, credential mismatch, timely filing). For a clear fix like a missing modifier or wrong place-of-service code, it prepares the corrected claim and routes it for one-click resubmit. It also surfaces denial patterns by payer and code so the team fixes the upstream cause, not just the one claim.
The AI works the queue; your billing coordinator approves every resubmission. It works on top of your clearinghouse, never replacing it.
Denials get worked before the appeal window closes, not after.
Medicaid and state-funded ABA programs increasingly require Electronic Visit Verification for every session. The right provider, client, location, and service type must match the state's EVV aggregator in real time. A session that does not match the EVV record denies, and the mismatch is not always visible until the claim comes back weeks later.
An AI employee reconciles every session entry against the EVV response (Sandata or your state's aggregator) before the claim goes out. It flags mismatches and routes the correction to a billing coordinator to clear. It also watches for Medicaid-plan coverage lapses between sessions and sends a same-day alert so the coordinator can re-verify before the next session runs.
A billing coordinator approves every correction and every coverage-lapse flag.
EVV mismatches get cleared before they become denials.
Inquiries arrive across four channels at once: website form, phone, referral fax from a school or pediatrician, sometimes a payer-portal message. After-hours calls hit voicemail and sit until the next business day. Slow intake hands the slot to whoever called the family back first.
An AI employee monitors every intake channel in real time. It screens each inquiry against the clinic's contracted payers and service areas, then sends a same-day response with an intake link, even after hours. It runs a 270/271 eligibility query the night before the first appointment, logs the response into the EHR, and flags coverage lapses before the visit instead of at the denial 45 days later. It chases incomplete intake packets and pre-fills what it already has.
A front-desk coordinator confirms the triage and finalizes the record.
Families get a same-day response instead of waiting until Monday.
ABA is delivered at high weekly frequency (10 to 40 hours per client across multiple RBTs and a supervising BCBA), so scheduling is a constraint problem. One cancellation cascades: the RBT's slot goes empty, an authorized hour goes undelivered, and that underutilization shows up as a unit reduction at the next re-auth.
An AI employee watches the live schedule against authorized units by CPT code. When a cancellation hits, it finds the waitlist families that match the open slot's constraints (payer, RBT credential, session type), drafts the fill-the-slot outreach, and surfaces the confirmed fill for a coordinator to approve.
A staff member approves every send and every booking. Scheduling respects session type against the authorization and payer-specific BCBA-to-RBT supervision ratios.
No-show recovery moves from reactive to proactive, and authorized hours stop going undelivered.
Every commercial payer requires a current, attested CAQH profile. Re-attestation comes due every 120 days per CAQH ProView documentation. A newly hired RBT or BCBA who starts seeing patients before their payer enrollment clears generates a billing blackout. A single missed OIG exclusion check can trigger a payer audit.
An AI employee monitors the credentialing matrix per provider, location, and payer: CAQH re-attestation deadlines, license expirations, payer-enrollment gaps, and OIG check windows. Thirty days before a profile lapses, it drafts the task and pre-fills the updated data. It gates a new hire out of the scheduling module until verification returns clean.
Your credentialing coordinator signs off on every renewal and every enrollment; the AI tracks and drafts. New-provider payer enrollment commonly runs 60 to 90 days, so the 30-day lead is load-bearing.
Claims stop rejecting because a credential expired silently.
ABA payroll is credential-sensitive and payer-dependent. RBT rates vary by session type, BCBAs are paid differently for supervising versus delivering direct services, and the session data lives in the EHR while the rate table lives in ADP, Gusto, or Paylocity. A staff member bridges the two by hand every pay period, and mismatches become either payroll errors or claim errors.
An AI employee runs a payroll-readiness check at the close of each period. It matches every session record to its time entry, flags duration mismatches, confirms the rate tier against the session type, and produces a reconciliation summary for the HR or billing coordinator to approve before the export runs.
The AI reconciles; a human approves before any payroll is cut. It works on top of the payroll system you already run, never replacing it.
Pay-period errors get caught before they become claim errors or staff complaints.
Every denial you don't work in time is permanent.
In 30 minutes we'll show you exactly where your ABA group is leaking revenue. Built by a former compliance officer at a multi-location pediatric therapy group.
See where you're leakingWorks on top of your stack.
Relay does not replace what you run. AI employees sit in the gaps between the systems you already have and hand each finalized action to your team.
CentralReach is the dominant ABA-native EHR/PM. Its REST API (OAuth 2.0/JWT, Enhanced APIs, Preferred Partner Network) exposes scheduling, authorization, billing, and clinical-note endpoints. AI employees read auth units, schedules, and claim status from CentralReach and surface the next action for your team. On top of CentralReach, never replacing it.
Motivity (clinical-first ABA-native) exposes a Workforce API (JSON REST and a Windows DLL) for bidirectional sync, plus CSV export and an ecosystem spanning clearinghouse, EVV, payroll, and CRM. AI employees use that surface to track authorizations, reconcile session notes against billing, and chase incomplete patient packets, with a staff member finalizing every action.
AlohaABA (billing-centric ABA) offers a REST API (OAuth 2.0) and connects to Availity/Office Ally clearinghouses, ADP/Paylocity/Paychex/Gusto payroll, and Sandata EVV. AI employees layer on top to connect authorization tracking to scheduling, work the EVV reconciliation loop, and work denials and AR that today require manual spreadsheet exports.
RethinkBH (ABA and multidisciplinary pediatric) has a Live API and a nightly Data Warehouse, plus Marketplace integrations. AI employees read its authorization and billing queues to flag expiring auths and unworked denials across locations, with your team approving every action.
For ABA platforms that run closed or near-closed (Theralytics, Noteable, Ensora ABA Suite/WebABA, Raven Health, AccuPoint), AI employees work through clearinghouse feeds (Office Ally, Claim.MD, Availity, Waystar), e-fax intake parsing inbound auth approvals by OCR, and exports, so the integration model holds even where there is no public API. Staff finalizes every action.
AI employees also work across the tools that ring the EHR: Availity and Waystar (eligibility, claims, denials), careviso and Cohere Health (eligibility-to-prior-auth), SRFax and Fax.Plus (reading inbound payer faxes), CAQH ProView, Modio Health, and Verifiable (credentialing), ADP, Gusto, and Paylocity (payroll), and Power BI (one operational dashboard). One view across all of it; your team finalizes.
Layer-on vs. rip-and-replace vs. hiring more staff.
ABA groups evaluating automation typically consider three paths. The right choice depends on how many of the failure points above you are hitting at once.
| Approach | What it means in practice | The tradeoff |
|---|---|---|
| Rip-and-replace (new EHR or platform) | Replaces the system your clinical team already knows; migration runs months, re-training runs longer | Automation built for the median clinic, not your payer mix; if the platform is acquired, you migrate again |
| Generic point tools (single-workflow chatbot or RPA bot) | Solves one queue without touching the adjacent workflows that feed or follow it | Auth expiration still leaks and denials still queue, because the tools rarely talk to each other |
| Hiring more billing staff | Scales linearly with location count | Cannot watch every auth, denial, and EVV queue simultaneously across all locations |
| Layer-on custom AI employees (Relay) | Works on top of the EHR you already run, built for your payer mix and specific failure points | A staff member finalizes every action; no rip-and-replace; recurring monthly fee |
Rip-and-replace (new EHR or platform)
What it means in practice
Replaces the system your clinical team already knows; migration runs months, re-training runs longer
The tradeoff
Automation built for the median clinic, not your payer mix; if the platform is acquired, you migrate again
Generic point tools (single-workflow chatbot or RPA bot)
What it means in practice
Solves one queue without touching the adjacent workflows that feed or follow it
The tradeoff
Auth expiration still leaks and denials still queue, because the tools rarely talk to each other
Hiring more billing staff
What it means in practice
Scales linearly with location count
The tradeoff
Cannot watch every auth, denial, and EVV queue simultaneously across all locations
Layer-on custom AI employees (Relay)
What it means in practice
Works on top of the EHR you already run, built for your payer mix and specific failure points
The tradeoff
A staff member finalizes every action; no rip-and-replace; recurring monthly fee
Automate the ABA clinic workflows first; switch the platform when you have a reason beyond the billing queue. Relay is human-in-the-loop by design, so a staff member at your clinic finalizes every claim and every appeal. The workflows are written for ABA and pediatric therapy, not adapted from a generic billing tool.
Already on Thoughtful AI and getting moved off it.
Thoughtful AI has redirected to smarterdx.com following a roll-up into Smarter Technologies. Clinics that ran their ABA billing on it are now looking for somewhere to land.
The dedicated migration guide covers the full picture. AI employees work on top of the EHR you already run, so keeping ABA billing moving does not mean switching your whole system. Your team finalizes every claim.
How we build it.
We start from the problem you feel, then build the fix on the systems you already run. Discovery and your first working AI employee take 2 to 3 weeks. The full build runs 8 to 12 weeks.
Start with a free 30-minute call
A short call about where the work is piling up and what that is costing you while it stays manual. No commitment, and you leave knowing where you would start.
Discovery and your first AI employee (weeks 1 to 3)
A few working sessions with your team. We map your operation end to end, every workflow across your locations, and find where the money leaks and what closing it is worth. You do not walk away with just a document. By the end of discovery we have built your first working AI employee on top of the systems you already run, so you see it pay off in your real setup before the full build starts.
The full build (8 to 12 weeks, start to finish)
We build the rest of the AI employees you mapped and wire them across every location. Nothing goes out until your team approves it, so you stay in control the whole way. One pediatric therapy client had all seven locations live within 90 days.
Proof: Sensory Speech & Occupational Therapy.
Sensory Speech and Occupational Therapy is a multi-location pediatric speech and OT group. Relay built two AI employees on top of their existing EHR and Drive, with a staff member finalizing every action.
The intake AI employee runs the full new-client lifecycle: schedules clinic tours, gets ROIs signed, requests records from schools and prior speech and OT clinics, requests IEPs, sends medical orders to the child's PCP and follows up until signed, starts authorization renewals about a month out, and sends three-month progress reports and evaluations to PCPs for signature.
The internal auditing AI employee reviews every clinical note nightly against the clinic's requirements, confirms the billing code matches the note, and after billing finds and appeals denied claims and reconciles remittances against the EHR notes.
The group saw 100% claim accuracy, staff 33% more productive, claim denials down 12%, and faster documentation turnaround. These are client-reported results for their operation; individual outcomes depend on your payer mix, volume, and stack.
AI employees for ABA clinics: frequently asked questions.
One view, live, for every client. An AI employee keeps a live auth tracker per client from the EHR you already run, flags units nearing expiration 30 days out, projects utilization against the schedule, and drafts the re-auth packet for a staff member to submit. The tracker covers every authorization type: direct therapy (97153), supervision (97155), and parent training (97156).
No. Relay is not an EHR. AI employees work on top of the ABA platform you already run, reading auth, scheduling, and claim data and surfacing the next action for your team to finalize. Nothing gets ripped out. The model holds whether your platform exposes a full API or runs through clearinghouse feeds and exports.
Yes, and they work the patterns too. They read every denial, classify it by reason code, draft the correction or appeal, and route it for one-click resubmit. They also surface repeating denial patterns by payer so you fix the upstream cause, not just the one claim. A billing coordinator approves each resubmission.
The common ABA family: 97151 (assessment), 97153 (direct therapy), 97155 (protocol modification), 97156 (parent/guardian training), and 97158 (group). The AI checks CPT and modifier combinations against payer-specific rules before a claim leaves your clearinghouse.
Your staff approves everything; the AI does the legwork. The AI employee handles the watching, checking, and drafting across billing, authorization, credentialing, and payroll. A staff member at your clinic reviews and finalizes every submission, send, and escalation. The AI never submits a claim, sends a message, or runs payroll on its own.
Keep billing moving without switching your whole system. Because AI employees work on top of your existing EHR rather than replacing it, you can keep ABA billing running on the platform you already have. The full picture is at the Thoughtful AI alternative page; your team finalizes every claim.
The fill happens before the slot goes cold. When a slot cancels, the AI matches waitlist families against the open slot's constraints (payer, RBT credential, session type, authorized units), drafts the outreach, and surfaces the confirmed fill for a coordinator to approve. No slot goes unworked; no outreach goes without a human sign-off.
Yes. The AI monitors CAQH re-attestation (due every 120 days per CAQH ProView documentation), license expirations, OIG checks, and payer-enrollment gaps per provider and location, and drafts renewals 30 days out so claims do not start rejecting silently.
Every session gets checked before the claim goes out. The AI employee reconciles each session entry against the EVV response (Sandata or your state's aggregator), flags mismatches, and watches for Medicaid coverage lapses between sessions. A billing coordinator clears every flagged item before the claim runs.
Relay builds for US-based multi-location healthcare clinics. A staff member finalizes every action, and access is scoped to the data each AI employee actually needs. The intro call covers exactly how it fits your stack and your compliance posture.
A free 30-minute intro call. We map where you're leaking; the first few weeks are discovery (working sessions, mapping the operation, building the first AI employee). Pricing is a recurring monthly fee. No one-time build, no installments.
Yes. Where a platform runs closed (for example Theralytics, Noteable, Ensora ABA Suite), AI employees work through clearinghouse feeds, e-fax parsing, and exports, so the model holds without an API.
Yes. The AI checks place-of-service against payer rules for in-home vs clinic sessions (a common denial cause) before claims go out.
See where your ABA group is leaking.
Every ABA location adds another auth tracker no one is watching, another denial queue no one has time to work, and another EVV mismatch no one catches until the claim denies. In 30 minutes we'll show you exactly where it's leaking and what to close first. It's a free intro call, no commitment.