Led the development of an agentic AI assistant that transformed Paradox's in-product help widget from routing-based responses into an intelligent, task-executing assistant.

Rigid routing-based responses forced users to complete tasks manually, creating friction across complex workflows. Analysis of 29,000+ interactions showed demand for automation and bulk actions the system couldn't support, leading to user drop-off, eroded trust, and growing competitive pressure.
Built an agentic AI assistant that executes complex workflows end-to-end via natural conversation, supporting scheduling, job and candidate management, and availability updates with secure permissions, approval flows, and real-time task visibility.
40% increase in adoption rate
30% increase in engagement depth
60-80% time savings on routine recruiting tasks while maintaining quality standards
Increased self-serve capabilities enabling users to complete complex operations without deep product knowledge