Project Overview
I led the end-to-end product management of a new LLM-powered natural language search interface in Cisco Observability, aimed at improving data discoverability and reducing user friction with complex query syntax. This initiative drove a 30% reduction in time-to-value and significantly increased user satisfaction and engagement.
• Role: Product Manager
• Team: 15 cross-functional members including data scientists, designers, engineers, researchers, QA, and domain PMs
• Team: 15 cross-functional members including data scientists, designers, engineers, researchers, QA, and domain PMs
Problem & Opportunity
Search is one of the most frequently used features in our platform, yet users struggled with the existing query-based filter interface:
• High error and drop-off rates due to the steep learning curve
• Significant time spent learning query syntax, resulting in low efficiency
• Frequent support requests and reliance on documentation for basic queries
This presented a clear opportunity to reimagine search with a natural language interface that would be more accessible and intuitive for all users.
Target Users & Insights
To deeply understand user needs, I partnered with a user researcher to conduct in-depth interviews with:
• End users, to uncover pain points and desired workflows
• Sales and support teams, to identify recurring usability challenges
• 10+ domain product managers, to align on platform-wide requirements and ensure scalability
I also conducted a competitive analysis of natural language and filtering experiences across observability and analytics tools.
Key insights:
• Users wanted a search experience that felt more conversational and intuitive
• There was a strong need for guidance when starting a query
• Advanced users still valued having access to the original query mode
Strategy & Roadmap
I defined the product vision and strategy for a dual-mode, LLM-powered search experience. My responsibilities included:
• Driving cross-functional alignment with engineering, design, data science, and other product teams
• Defining the MVP scope, prioritizing high-impact workflows
• Building a phased roadmap to enable early delivery and iterative testing under tight timelines
UX Design & Testing
I worked closely with design to turn requirements into interactive prototypes and led usability testing to validate assumptions and iterate based on feedback. I also collaborated with data scientists to shape the model behavior and ensure alignment with user expectations.
Core Features Delivered
• LLM-Powered Natural Language Interface: Enables users to search data using plain English, eliminating the need to learn complex syntax
• Text Prompts: Contextual sample queries to help users get started faster
• Regenerate Button: Allows users to cycle through alternative query suggestions
• Dual Mode Search: Seamless toggle between natural language and classic query-based search to support different user personas
Results & Impact
• 30% reduction in time-to-value for users searching and exploring data
• Increased user engagement and satisfaction, validated by follow-up interviews
• Solution is now being scaled across additional workflows within Cisco Observability
