Context and Problem
Healthcare practices lose revenue and spend hours on manual billing readiness because Medicare and CMS rules are complex, change frequently, and require audit-ready documentation. Small errors like wrong CPT codes, missing time thresholds, and undocumented patient consent trigger billing errors and denials or compliance risk.
[Company], a chronic care management platform, supported 20+ billable care activity codes across multiple CMS programs (CCM, RPM, PCM). Clients were using spreadsheets and manual checks to validate billing readiness, which meant:
- Manual workload equivalent to ~2 FTEs per client spent on claim prep and validation
- High error rates leading to denials and rework
- No real-time visibility into billable vs non-billable activities
- Audit anxiety, practices could not prove compliance if questioned
Users and Constraints
Primary users
- Practice administrators (preparing claims, managing billing workflows)
- Care coordinators (documenting interactions, meeting time thresholds)
- Compliance officers (auditing records, preparing for reviews)
Constraints
- Integrations with major EMR and EHR systems, including Epic-based workflows
- CMS rules change annually, system needed configurability without engineering rewrites
- HIPAA-compliant data handling required
- Clients ranged from 2-person practices to 50+ care coordinator teams
What I Shipped
CPT Code Rules Engine
- Configured logic for 20+ care activity codes (time thresholds, consent requirements, documentation standards)
- Real-time validation flagged incomplete or non-compliant activities before billing prep
- Annual rule updates managed via an admin interface (no code deploy required)
Billing Readiness Workflow
- Aggregated billable activities by patient, provider, and billing period
- Generated billing-ready documentation with required details attached
- Supported streamlined handoff into practice management billing processes
Billable Minutes Dashboard
- Real-time view of billable vs non-billable time by care coordinator
- Alerts when patients were close to thresholds (example: 15 more minutes needed)
- Forecast view showing projected monthly revenue based on current activity
Audit Trail and Compliance Reports
- Logged patient interactions with timestamp, user, and activity type
- One-click audit reports for reviews (filterable by date range, provider, code)
- Automated checks flagged missing consent forms or documentation gaps
EMR and EHR Integrations
- Built data sync pipelines with major systems, including Epic-based workflows
- Mapped demographics, care team assignments, and interaction logs to EHR data models
- Handled edge cases like duplicates, partial syncs, and API limits
How AI Showed Up
AI was used as an internal accelerator for analysis, workflow design, and ops tooling, with strict constraints around compliance and human review. Where we prototyped AI-assisted documentation, outputs were draft-only and required human confirmation before being saved.
- All AI-generated content flagged as draft and required human review
- No automated billing decisions based on AI suggestions (approval required)
- Audit logs captured when AI was used vs when humans made edits
Outcomes
- Reduced manual workload equivalent to ~2 FTEs per client by eliminating repetitive billing prep tasks
- [X]% reduction in billing errors and denials due to real-time compliance validation
- [X]% increase in captured billable revenue as care teams identified previously missed activities
- 100% audit pass rate for clients using the system during [timeline]
- 5 client dashboards launched tracking utilization, revenue, and compliance metrics
My Role
What I owned
- Product strategy and roadmap, prioritized by revenue impact and compliance risk
- Requirements definition (PRDs with CPT logic, workflows, edge cases)
- SQL analysis to validate thresholds and identify revenue leakage
- Client feedback loops with administrators and care coordinators
- Integration specs (data model mapping, API requirements, vendor coordination)
What I collaborated on
- Engineering on rules engine architecture and pipeline design
- Compliance advisors to validate CMS rule interpretation
- Client success on onboarding workflows and training materials
What I Learned and What I Would Do Next
- Configurability matters: rules change, products must update without waiting on sprints.
- Real-time feedback wins: alerts and thresholds drive behavior better than perfect reports.
- Integrations are underestimated: one EHR is manageable, multiple requires normalization and strong data validation.
Next: predictive scheduling for outreach timing, denial pattern analysis to surface corrective actions, and cross-program optimization for eligible patients.