A Documentation Workflow Built for Qualifacts Credible
Clinical documentation inside Qualifacts Credible rarely fails because clinicians lack dedication. It fails because the platform is service-centric. Every partial hospitalization program (PHP) participant generates a distinct service entry. Each entry needs its own note tied to the client's treatment plan, DSM-5-TR problem list, and billing record. A single four-hour PHP day for a full cohort can produce dozens of service notes before the clinician leaves the building.
Merry AI addresses this load by capturing the group session once and processing the audio in volatile memory. It then splits the transcript into individualized per-client drafts. Each draft reflects that client's participation, affect, risk indicators, and response to intervention rather than a cloned group narrative. The clinician reviews every draft, then commits it to the correct Credible service entry in the browser already open on the workstation.
How the Chrome Extension Writes Into Credible
Credible is a browser-rendered, form-driven platform with configurable templates, calculated fields, and agency-specific layouts. Qualifacts maintains a backend API, but Merry AI does not depend on it. It does not claim native FHIR write access, and it does not route notes through a marketplace connector. Instead, the extension operates at the presentation layer, where clinicians already document. It writes text into the same input controls a clinician would type into, with no marketplace fees and no API lag.
Service Entry Launch and Client Mapping
The workflow begins when the clinician launches a Credible service entry in Chrome. The extension detects the documentation screen through URL patterns and DOM cues such as section headers and field labels. It reads the client name rendered on the service entry and maps that name to the matching per-client draft. Merry populates the note field only after the clinician confirms the match. When multiple service entries are open in separate tabs, each tab is bound to one client, which prevents cross-chart contamination.
Write-Only Injection Boundaries
Injection is deliberately narrow and write-only. Merry targets designated textarea and rich-text controls such as interventions, client response, and plan. It never modifies hidden metadata fields that hold service, client, or staff identifiers. Because Credible forms are agency-configurable, selector maps are maintained per deployment and updated when Qualifacts releases interface changes. Native input events fire after each write, so Credible's validation and calculated fields behave as they would with keyboard entry.
The architectural tradeoffs are summarized below for community behavioral health organizations comparing documentation options.
| Architecture | Deployment Overhead | Latency | Annual Fee | Per-Client Group Splitting |
|---|---|---|---|---|
| Traditional API Marketplace Integration | Vendor contracting, IT security review, interface build | Queue- and sync-dependent | Recurring marketplace and connector fees | Depends on vendor build |
| Manual Copy-Paste | None | Minutes per note, high transcription error risk | None, but heavy labor cost | Manual duplication and editing |
| Merry AI Chrome DOM Injection | Chrome extension deployment, no firewall changes | Under 500ms write to the open field | No marketplace fees | Automated, clinician-confirmed |
PHP Block Documentation Integrity
Partial hospitalization documentation is audited against a higher bar than routine outpatient therapy. Reviewers expect each note to show that PHP intensity remains medically necessary. They also expect group interventions to target specific DSM-5-TR symptoms. Finally, the client's individual response must be recorded rather than inferred from attendance.
Start Time, Stop Time, and Modality
Every PHP block Merry drafts records start time, stop time, total duration, and group modality, such as CBT skills, DBT distress tolerance, psychoeducation, or relapse prevention. These times must match the Credible service entry metadata, so the clinician verifies them before signature. Each block then links the intervention to an individualized treatment plan goal. It states what the client said or did, how that response relates to the goal, and whether progress, stasis, or regression occurred.
Individualization is the audit defense. Two clients attending the same DBT block receive notes with shared group content but distinct observations, risk screening results, and plan entries. Clinicians can browse PHP, IOP, and group therapy structures at https://templates.scribing.io and adapt them to their agency's Credible form build.
Compliance, Retention, and Audit Defense
Billing integrity depends on separation. PHP group services are not evaluation and management visits, so HCPCS G2211 never attaches to a group note. When a client also receives a same-day psychiatric E/M visit, Merry generates a separate E/M note with its own history, assessment, and medical decision-making. That note supports modifier 25 where the criteria are met. CMS does not pay G2211 on an E/M reported with modifier 25 except alongside specified preventive services. Where modifier 59 is contemplated under NCCI policy, distinct times and clinical purposes are documented explicitly rather than assumed.
Raw audio never persists beyond RAM. No audio file is written to disk or retained in vendor storage. This narrows breach exposure and aligns with data minimization expectations for programs subject to 42 CFR Part 2. Consent and AI-use disclosure workflows remain the agency's responsibility. Every draft stays a draft until the clinician reviews, edits, and signs it inside Credible.
Evidence on documentation burden in behavioral health continues to accumulate in the peer-reviewed literature indexed at https://www.ncbi.nlm.nih.gov/pmc/. Agencies evaluating Merry AI should measure time-to-signature, note individualization, and claim denial rates against their own baseline before and after deployment.
