The problem

MEDsights captured insights, but the workflow immediately left the product. Teams exported hundreds of records to Excel to review, compare, group, assign ownership, and track outcomes—leaving adoption at 14% with no reliable traceability from insight to impact.

The strategic shift

Research across 19 interviews and 12 countries showed that the problem was not a feature gap—it was a handoff gap. Instead of building the requested AI chat experience, I reframed MEDsights around the complete insight-to-outcome workflow and defined where AI should assist versus where human accountability had to remain.

Key decisions

• End-to-end workflow model replacing five spreadsheet-based handoffs

• AI-assisted grouping, summarization, and trend detection

• Human-review model for validating and publishing AI recommendations

• Dashboard, insight, trend, ownership, and outcome-tracking experiences

• MVP prioritization that cut AI chat while shipping 12 higher-value workflow capabilities

Outcome

Five of six workflow steps moved back inside MEDsights while the consequential medical decision deliberately remained human. Within three months, adoption increased from 14% to 26%, exceeding the 24% target; onboarding and training costs decreased 14%; and teams across 12 countries reported better alignment.