In late 2023, California's $4B Children and Youth Behavioral Health Initiative needed something the state had never built before: a way for school districts to deliver — and bill for — Medicaid-covered behavioral health services at scale. This is what it took to build the infrastructure underneath that ambition.
The context.
California's Children and Youth Behavioral Health Initiative is one of the largest public investments in youth mental health in US history. The program's premise is simple in language and unprecedented in operations: meet kids where they are — at school — and pay providers to serve them through the same Medicaid plumbing that funds hospitals and clinics.
The trouble is that schools are not healthcare providers. Local Education Agencies — county offices, school districts, charter networks — do not have Electronic Health Records, Charge Description Masters, payer enrollment apparatus, claim-scrubbing logic, or denial-management teams. They have student information systems and instructional staff. The gap between those two worlds is the problem this engagement existed to close.
What had to be built.
The mandate covered ten Local Education Agencies serving roughly 120,000 students, with the obligation to bill cleanly against the CYBHI fee schedule, the Medi-Cal Billing Option Payment program, and the School-based Medi-Cal Administrative Activities (SMAA) program. There was no precedent system to copy. The build had to be done while the field of providers — counselors, social workers, behavioral health specialists — was still being hired.
The work fell into four layers, designed and stood up in parallel:
Designed from a blueprint covering system architecture, UI/UX, user roles, and business logic. The system supports providers, clinical supervisors, billing coordinators, administrators, and LEA business officials — each with role-based dashboards calibrated to what they actually need to see. Seven dashboards in total, each born from a real conversation about what the work looks like.
Co-designed the data hierarchy, ETL pipelines, and the analytics layer feeding executive and operational dashboards. The warehouse handles claims, eligibility, encounter, and reimbursement data — and was deliberately built so that the same numbers would show up the same way whether a county business official, a state auditor, or a clinical supervisor was looking.
Established a Charge Description Master, claims-scrubbing logic, reimbursement-rate tables, and full 837P billing workflows for three different programs simultaneously: the CYBHI multi-payer fee schedule, Medi-Cal BOP, and SMAA. Each program has its own rules, its own forms, its own gotchas — and providers do not have the luxury of caring which is which. The system has to.
No EHR survives contact with its users unless someone is on the other end of the help line. Built the helpdesk function, the training and FAQ library, the user manuals — and instituted a standing enhancement-and-bug-fix governance process so the system keeps getting better instead of slowly worse.
The system, in brief.
- EHR Platform Salesforce Health Cloud (custom build)
- Data Warehouse Azure SQL with custom ETL pipelines
- Claims Submission 837P via Availity clearinghouse
- Payment Reconciliation Carelon TPA and Medi-Cal
- Programs Covered CYBHI fee schedule · Medi-Cal BOP · SMAA
- Compliance HIPAA · Medi-Cal · No Surprises Act
- Coverage 10 Local Education Agencies · 120,000+ students
- Team 15-person multi-disciplinary group across billing, claims, provider data, helpdesk
What the numbers say.
The numbers above are presented for what they are: directional indicators of a system that works, not the final measure of it. A 90% clean-claim rate is a meaningful threshold for a Medi-Cal billing operation — particularly one running against three programs simultaneously, on a platform that did not exist two years ago.
The other measure — harder to quantify but more important — is that ten LEAs now have a way to fund the people doing the work. A district that hires a counselor knows the claim will go through. A county business official can answer the question "did we get paid for that?" without a forensic exercise. That is what the system is for.
What it taught me.
The work continues. The system gets better. The next chapters — adding LEAs, expanding programs, integrating more deeply with the state's data infrastructure — are still being written.