ELI LILLY
Integrating AI to support the full lifecycle of healthcare market research
Context
Since our inception, we have advanced organizations across sectors by pairing strategic insight with disciplined execution.Our multidisciplinary team analyzes each objective, designs a tailored roadmap, and implements with rigorous oversight.
Challenge
Lilly asked us to elevate the MVP of Perspective through data expansion and a ground-up redesign guided by user research, while intentionally keeping the human in the loop.
Discovery and Research
Through a set of structured interviews aimed at understanding the role of Perspective in market researchers day-to-day workflows, we identified key insights and themes to guide the redesign.
Unstructured Knowledge Network
Power-users developed their own methods for tool use and served as thought leaders and mentors to their teammates—creating an informal knowledge network surrounding the tool.
Unexpected Use Cases
Researchers were using Perspective to plan research, not just run it. They asked personas how they’d like to be approached and used the knowledge base as a context engine to define knowns and unknowns.
Synthesis Friction and Disconnect
Perspective generated quality insights, but extracting them was a friction point. Synthesis happened outside the research environment, in Claude, which led to lost context and occasional hallucinations.
Key Takeaway
Researchers take pride in their instincts, expertise, and craft. The rebuild needed to sharpen their capabilities, not replace their judgment.
The Solution
The research led us to design a two-mode conversational setup, complemented by spaces for sharing prompting methods and insights across teams.
Prompt Library
The Prompt Library gives structure to Lilly’s peer-to-peer learning culture. Users can publish their own methods and best practices for broader adoption.
Insight Library
The Insight Library removes silos around insights generated in Perspective by giving users a place to publish valuable findings.
Interview Space
In the Interview Space, users can query one or more synthetic personas using prompts from the Prompt Library, a discussion guide generated in the Companion Space, or freeform questions.
When valuable insights surface, users can save them flexibly: highlight a single line, save one response, or save every persona’s response at once.
Companion Space
In the Companion Space, users synthesize research into an editable .docx through a user-guided, AI-supported workflow, turning findings into structured, shareable artifacts.
How We Built It
I took an unconventional path, vibe-coding the prototype front to back and working across Cursor and Figma to define design systems, UI patterns, and a source of truth.
Our data and technology strategist built out the tool in parallel as the design took shape, which let us release in a tight three-month timeline.
~$4.9M
Costs saved since launch
2,180
Hours saved since launch
26,000
Responses generated
66
Primary research projects avoided
Outcome