INNOVACCER / PATIENT 360
OVERVIEW
When a doctor sees a patient, they document diagnoses, prescriptions, and clinical notes. A medical coder then reviews these records to check if everything was captured – because what's not documented means lost funding and incomplete care history.
The problem: the doctor and the coder use different tools. The doctor documents in InNote during the visit. The coder reviews in Patient 360 after. Neither has complete visibility into what the other has done.
My challenge: connect these two workflows so the coder's review is informed by what the doctor already documented closing the loop across both products.
MY ROLE
I owned Patient 360's (product housing patient records) risk adjustment experience – driving improvements across the workflow over time and adding new features like an AI Agent, query template management, provider response visibility, and more – as I cross-collaborated with the InNote product and design team.
IMPACT
Review time reduced from ~15 min to under 5 min for 79% of users. 146K+ coding gaps reviewed across 30+ health systems.
TEAM
2 Senior product managers, Engineering team · Cross-collaboration with the InNote product and design team
What the healthcare environment looks like..
Healthcare runs on three players. Providers, Patients & Payers (insurance). The link between all three is documentation (prescriptions, visit notes, discharge summaries, diagnoses, etc) - what the doctor writes down during a visit.
Two tools. Same patient. No shared view.
Doctors document what a patient has – eg. diabetes, heart failure, COPD – into an EHR (Electronic Health Record) during visits. Insurance reimburses hospitals based on what's documented – if a condition isn't in the record, the hospital doesn't get paid for treating it.
P360 surfaces conditions a patient is known to have or is suspected to have and represents them as standardised codes. A medical coder reviews these conditions and decides - approve, or reject based on the evidences available from the documentation.
But the doctor works in a separate app called InNote. What they've already documented during the visit never made it into the coder's view.
How might we bring the doctor's decisions and documentation into the coder's review, without either of them switching tools?
The workflow existed. Just not in the product.
4 sessions with 3 coders, customer calls, SME interviews, and a Mixpanel audit gave us a clear picture.
From blind spot to full picture
AFTER VISIT
What challenged and shaped me
Cross-team collaboration: I owned P360, another designer owned InNote. Every feature touched both systems - which meant aligning with stakeholders across two product and engineering teams. We ran joint sessions, working through workflows, edge cases, and status mappings together instead of through handoff docs.
AI needs to build trust. In a domain like healthcare, fully automating clinical decisions isn't an option, human judgment is essential. Coders were uncomfortable with AI making decisions on codes, so we repositioned our Agent as a reviewer that assisted them instead.
Changing the IA without breaking habits. Years of muscle memory around billing codes and workflow meant transitioning in a way without changing the structure too much. We validated the new information architecture with SMEs across multiple rounds, ensuring the transition felt familiar even as the underlying organisation changed.
Constraints shaped the product. Decisions like limiting queries to one per code, making templates mandatory, and defining task triggers weren't handed to me - they came out of research and design exploration, and I helped shape them alongside the PMs.
Stakeholder mapping is crucial before starting. Identifying and aligning with the right people early - SMEs, users, PMs, engineering leads - set the foundation for our research, validation rounds, and feedback loops throughout the project.
Shipped and adopted at scale
Adopted by 30+ customers
