Experiment Memory Agent
An AI assistant to turn fragmented test archives into reusable decision intelligence.
The Opportunity
Experiment learnings were spread across tickets, comments, screenshots, and PDFs. Teams struggled to answer basic questions quickly, like whether a similar test had already been run and what it taught us.
The Approach
I built an AI-powered assistant that retrieves historical experiment metadata, summarizes outcomes, and surfaces reusable patterns through natural-language queries to support faster hypothesis validation.
Outcomes
- Reduced time spent searching historical experiments
- Improved knowledge reuse across the tribe and horizontal teams
- Increased experimentation velocity by shortening validation cycles
- Demonstrated practical AI integration in daily product workflows
Demo
Tools Used
Claude (Bedrock)
Python
API Integration