Custom Patient App, AI Care Companion, and Revenue Cycle Automation for Cardiovascular Disease Management

Insurance enrolment automation. Went live for the primary PBM carriers. The median referral to active patient window dropped from several days to under 48 hours.

The second of the two primary PBM carriers presented an unexpected obstacle. Discovery had mapped the carrier’s public API documentation, but the actual enrolment confirmation flow required an undocumented callback verification step the documentation did not mention. The team spent four weeks working directly with the carrier’s developer relations group to obtain the verification flow and integrate it. The delay pushed the second PBM carrier live a month after the first, and the integration approach uncovered during the work later accelerated integration with two additional secondary carriers in Phase 4c.

Automated partner reporting. Replaced the manual process across all hospital partners, standardising scheduled monthly reports to a single cadence. The self service partner portal followed in Phase 4b, giving hospitals on demand access to the same data.

AI session intelligence. Deployed with both functions. Practitioners received real time protocol suggestions during sessions through rolling transcript analysis, and management received automated compliance reports after each session, replacing the manual review process.

Two weeks into the pilot, the clinical team flagged that the AI care companion’s responses to lifestyle questions were drifting toward general DASH diet guidance rather than the Client’s proprietary sodium reduction schedule. The retrieval index was treating the Client’s protocol and external clinical reference material with similar weight, so the model occasionally pulled in generic advice.

“It would tell a patient to gradually reduce sodium when our protocol calls for a specific four week step down,”

~ One of the Client’s senior clinicians explained during the review.

The team rebuilt the retrieval index with explicit document priority weighting, placing the Client’s protocol documentation at the top of the retrieval hierarchy and treating external reference material only as a fallback when no protocol guidance applied. The pilot resumed two weeks later, with the clinical team approving the revised responses across a sample of forty patient queries.