Showcase
Designing AI-Driven solutions for sales efficiency
This is a high-level showcase of my case study. Want to dive deeper into the process? Click here
My Role
Conducted and integrated UX research to redefine the initial problem and inform AI model development alongside AI/ML engineers.
Collaborated with ML engineers to collect ground truth data and advocated for the F1 score as a key success metric.
Introduced AI components to the design system to strengthen our AI value proposition and enhance transparency by clearly signaling AI-driven features to users
Partnered with the PM lead to define success metrics and align them with OKRs.
Collaborated with the PM lead to define the AI-driven solutions roadmap, with this project serving as the foundation for future developments.
Impact
Customers
Reducing response time—the time it takes for sales teams to reply to high-intent leads—from 3 days to 1 day led to more booked meetings, confirming that faster engagement with qualified prospects improves conversion rates.
Business
By providing data visibility into AI performance, this project enables Copilot to track and validate lead quality, strengthening trust in AI recommendations and reinforcing its market differentiation.
Solution
Based on my research (deep dive here), we found that Sales teams prioritize leads based on intent. To support this, we developed an AI model that predicts lead intent from conversation threads, following the same intent categories uncovered during research. Since highly qualified leads often express interest early, we introduced a new inbox tab to surface them—ensuring they stood out in the workflow, not just in reports.
We introduced a dual-label system to display conversation intent analysis. The design includes a feedback mechanism to improve model training while addressing customers' need for accurate reporting
When labels are manually selected, AI suggestions remain available, allowing customers to review and correct any mistakes if needed.