Clinical documentation, audited at scale
BH QA Sentinel is the documentation and revenue infrastructure for behavioral health and SUD treatment programs. Eighteen agents read your EMR, audit every clinical note against 54 rules, and file the fixes before the payer ever sees the gap.
| Clinician Name | Missed Action Hours | Number of Clients |
|---|---|---|
DEDr. Emily Rodriguez | 18 | 10 |
DMDr. Michael Chen | 16 | 38 |
SKSarah Kim, LCSW | 15 | 12 |
JPJames Patel, MFT | 14 | 16 |
RCRobin Cole, RN | 9 | 10 |
One platform that closes the loop from note written to claim paid.
Most documentation tools just audit. We audit, route, file the fix, and brief your team in the morning. The agents live inside your EMR. No data leaves your tenancy.
Every clinical note, against 54 rules, on every shift.
Audit Simulation runs every two hours over plans, sessions, and client records. Medical necessity, signature integrity, CPT-ICD alignment, ASAM/LOC coverage, group individualization, 42 CFR Part 2 redisclosure language.
{
rule: "MN-05",
client: "—",
finding: "session note not tied to a plan goal",
severity: "moderate",
amount_at_risk: 312.50,
suggested_fix: "link session goal to TP-007 (Anxiety)",
deep_link: "https://app.alleva.io/sessions/82441/edit"
}The right clinician gets the right nudge at the right hour.
Unsigned-note nagger escalates at 7:15, 12:15, and 17:15. After 24 hours open, supervisor escalation fires. Each clinician sees their personal inbox, deep-linked to the chart, ordered by dollar exposure.
When you authorize the fix, the agent writes it back.
Ten write methods, API-first. Sign treatment plans, patch session fields, add CIWA/COWS, append addendum notes. Where the EMR API stops, a headless-Chromium worker logs in with PIN signing and edits the note body through the UI.
POST /alleva/v1/treatment-plans/4421/sign
Authorization: Bearer ${TENANT_TOKEN}
Content-Type: application/json
{
"clinician_id": 188,
"signature_date": "2026-05-14T07:12:04Z",
"pin": "${SIGNED_PIN}",
"audit_ref": "bhqa-finding-9a7c1"
}
→ 200 OK plan locked, downstream claim clearedAn always-on clinical operations team, written in code.
Runs the full 54-rule catalog over plans, client records, and session notes.
Pre-submission clean-claim checks. CPT, modifiers, units, dx alignment.
Escalating reminders for unsigned session notes, routed by hours overdue.
Surfaces plans whose endDate has passed without re-sign, ranked by exposure.
Flags long-stay actives and identifies clients who may be discharge-ready.
Anything still open past 24 hours escalates to the clinical supervisor.
Applies queued chart-fix proposals when writes are enabled on your tenant.
The morning brief, active clients, expired plans, readmits, LOC mix.
The math is the brochure.
“The morning brief is the first thing I open. By 8 AM I know which of my clinicians missed a signature, which plans are about to expire, and which claims are waiting on a fix the agent already queued. Nothing else got close.”
A two-hour install. Zero EMR migration.
The agent reads your EMR through the API your vendor already exposes. No template rip-and-replace. No data warehouse. No middleware vendor. We point at your tenant, sign the BAA, and run.
- 01BAA + tenant credentialsStandard subprocessor chain, your counsel signs nothing surprising.
- 02Connection verifiedPull a test client, confirm structure, confirm permissions.
- 03Rule pack tunedPick what fires, what auto-fixes, who gets routed.
- 04First morning briefBy 7 AM Wednesday, your director has the digest.
# alleva tenant
ALLEVA_TENANT_URL = "https://app.alleva.io/found-recovery"
ALLEVA_CLIENT_ID = "${TENANT_CLIENT_ID}"
ALLEVA_SECRET = "${TENANT_SECRET}"
# rule pack
RULES_ACTIVE = ["TP-*", "MN-05", "SIG-*", "GRP-*",
"ASAM-*", "QUAL-09", "REG-03"]
AUTO_FIX_ALLOWED = ["SIG-03", "QUAL-09", "MN-05"]
# routing
DIGEST_RECIPIENTS = ["director@clinic.com"]
ESCALATE_AFTER_HR = 24
$ bhqa-engine start
→ connected to Alleva (Found Recovery)
→ 18 agents scheduled
→ first digest queued for 07:00 PTBuilt for the audit your compliance officer is dreading.
BAA executed with every customer and every subprocessor in the chain. Per-tenant database, per-tenant encryption keys, per-tenant audit log.
SUD records get the federal handling they require. Redisclosure language enforced at the rule level (REG-03).
Flat-fee subscription pricing. No per-admit, no per-referral, no revenue share. The pricing structure is the product.
Type II audit in progress, target close Q3 2026. Controls already operating, not just designed.
Stop writing off the claims you already earned.
A 30 minute walkthrough, your data, your EMR. If we can not show you the gap in the first call, we will not waste your second.