How to Automate CentralReach

CentralReach keeps the count an ABA practice runs on

CentralReach is the electronic medical record and practice management platform used by ABA and multidisciplinary therapy providers. Its clients deliver applied behaviour analysis to children with autism and to people with intellectual and developmental disabilities, through registered behaviour technicians working under board certified analysts. Thousands of practices run on it, from a two-clinician startup to a multi-state provider.

A practice of this kind has an unusual shape. Care is delivered in units that a funder authorised in advance, usually Medicaid or a commercial payer, against a treatment plan somebody wrote and somebody else approved. A session produces data, then a note. Supervision carries its own required hours. Billing happens per unit, per service code, per authorisation.

So the whole business is a set of counts that have to agree: units authorised against units delivered, sessions against notes, notes against claims, hours worked against hours supervised. CentralReach is where those counts live.

The authorisation is a clock nobody in the clinic controls

Every one of those counts has a deadline attached, and the deadlines that matter most belong to somebody else.

An authorisation expires on a date the funder set. Renewing it means a reassessment, a report, a submission on that funder's portal, then somebody logging in every few days to see whether an answer appeared. Meanwhile the schedule keeps consuming units, and if they run out mid-month the sessions stop or get delivered unpaid.

So somebody, often a billing lead or a practice manager, spends the week like this. Check remaining units per client. Check which authorisations expire in the next two months. Check the funder portal for decisions. Check whether a technician's certification is still current on the register that governs it. Pull the denials and work them before the appeal window shuts.

It is clerical work with clinical consequences. A missed renewal is a child whose therapy pauses, and the family finds out on the morning it should have happened.

CentralReach

Renewals started early Authorisations approaching expiry surfaced with the units left on them, while a reassessment can still be booked.

Funder decisions found the day they land Portals checked each morning, rather than on the day a technician arrives and there are no units left to bill.

Credentials checked against the register Certification and enrolment dates read from the bodies that hold them, before a supervision hour becomes unbillable.

Every note gets audited before it becomes a claim

CentralReach has invested heavily in automating the parts that live inside it, and a practice that has not switched these on should start there.

CR NoteDraftAI drafts a session note from the data collected during the session, which makes the technician an editor and reviewer rather than an author. CR NoteGuardAI audits clinical notes, quarantines the ones carrying errors and assigns follow-up tasks to staff. CR ClaimCheckAI categorises, validates and organises claims, with filtering and audit logs so an administrator can label and act on them in bulk. CR ScheduleAI works on the schedule itself, and a dashboard tracks unit-based authorisations, showing total units, units used and utilisation rates. Around all of that sit data collection, assessments and curriculum, caregiver training and managed billing.

It is a lot, and it covers the practice's own records well.

Every piece of it, though, reads what CentralReach holds. The funder's decision is not in CentralReach until a person types it there. Neither is the reason a claim was rejected at the clearinghouse, the certification status on the certifying body's register, the state's provider enrolment record, or the school district's contract portal.

The funder's portal can be read before the first session

The counts inside the system are accurate. What is missing is the other half of each one, and it sits on a website somebody has to open.

Every open authorisation checked on the funder's portal each morning, with the decision, the approved units and the expiry date brought back and set against the schedule about to consume them. A renewal raised on the date the reassessment needs, rather than the week the units run dry.

Denials and rejections collected from payer portals and the clearinghouse, with reason codes and appeal deadlines attached, sorted by the date each one closes. Eligibility rechecked at the start of the month, because a family's Medicaid coverage can lapse without anyone in the clinic being told. Credentials checked against the register that actually governs them, before a supervision hour turns out to be unbillable.

This is clinical care, and that decides where the work stops. Gathering runs unattended, on a schedule. Anything that changes a clinical or billing record, or that reaches a family, is prepared and waits for a qualified person to review and send. The fetching goes away. The judgement does not.

The count can be kept without anyone counting

The authorised hours live on the funder's site and nowhere else, so reading them back is a login, a search and a note. In most practices that hour belongs to a supervisor, a clinician or a billing lead.

WebRun is an AI agent that works a real Chrome browser, signed in as your staff are. It opens the payer portal, the state system, the clearinghouse or the certification register, reads what is there, and puts it next to the client, the authorisation or the claim it belongs to.

It runs on your schedule in your own private environment, sessions are not shared between tools, and a workflow can be locked to an explicit list of sites. You can watch a run and stop it.

The workflows below are already built, and each one names exactly what it opens.