prompt engineers present
FALLING INTO CLEAN CODE
Automating Feature Flag Cleanup with AI
What are feature flags?
- Feature flags - enable/disable without deploying new code.
- Dynamic control and risk mitigation
- LaunchDarkly - UI platform that helps to toggle feature flags
694
flags in prod for over 60 days
471
flags in ON in production for over 60 days
178
the facts
flags that have been ON in Production for over 60 days and created before January 2025
8+
hours spent on cleanup
8 out of 9 teams spend 4 or more hours per sprint on feature flag cleanup
survey snapshot
Most teams conduct feature control activity every single sprint.
what are teams saying?
[Coordinating FF values across environments] for reporting needs time and a lot of co-ordination.
"...the major issue we have is finding time to remove flags when we have product roadmaps to keep up with. Our PO is going to allocate time on a per sprint basis starting next year to remove them."
...removal of code [around feature flags] needs more analysis on risk and testing work flows
- Improves Performance
- Reduces Network Load
- Optimizes Infrastructure Usage
- Controls Monitoring Costs
- Prevents Monitoring Gaps
why is cleanup important?
Technical debt Performance risk
Risk of errors Developer productivity
Over time, our system accumulates outdated LaunchDarkly feature flags that clutter the codebase, increase maintenance overhead, and pose risks to stability.
PROBLEM STATEMENT
proposed Solution
We propose an AI-powered solution to automatically identify and remove obsolete flags, improving efficiency and reducing technical debt.
AI-Driven Development Backend was designed and built using AI tools like Windsurf and GPT-4, accelerating prototyping and reducing manual effort.
Smart Orchestration Layer Java Spring Boot backend integrates with LaunchDarkly to serve retirement-ready flags and trigger intelligent automation.
Backend Approach
Seamless Integration - Backend applies AI-generated changes, commits them, and creates ready-to-review pull requests with detailed modification context.
Automated Refactoring GPT-4 performs deep code analysis and generates production-ready changes to safely remove feature flags across Java, JSP, and JS.
demo
- Frontend is stable
- Migrate backend logic to Jenkins pipeline
- Wrap frontend into webjar (similar to existing projects)
- DB table to track created PRs
- Overall development estimate ~1 month with 1-2 developers
Integration
Intelligent Flag Identification
Risk Aware Recommendations
Automated suggestions integrated to propose removals
Conclusions
Time and Effort Reduction
Cross-Team Visibility
Thank you!
'A clean codebase is like a crisp autumn morning—refreshing and productive.'
- COPILOT
Prompt Engineers
Niya anil Yashashree Majalikar Ganesh Ravendranathan
Saba Saba Saiprathap sankineni kishan trivedi
TECH STACK
Spring Boot
Angular
Azure AI Foundry
Windsurf
Demo screenshots
FALLING INTO CLEAN CODE
Yash Majalikar
Created on October 8, 2025
Start designing with a free template
Discover more than 1500 professional designs like these:
View
Smart Presentation
View
Practical Presentation
View
Essential Presentation
View
Akihabara Presentation
View
Flow Presentation
View
Dynamic Visual Presentation
View
Pastel Color Presentation
Explore all templates
Transcript
prompt engineers present
FALLING INTO CLEAN CODE
Automating Feature Flag Cleanup with AI
What are feature flags?
694
flags in prod for over 60 days
471
flags in ON in production for over 60 days
178
the facts
flags that have been ON in Production for over 60 days and created before January 2025
8+
hours spent on cleanup
8 out of 9 teams spend 4 or more hours per sprint on feature flag cleanup
survey snapshot
Most teams conduct feature control activity every single sprint.
what are teams saying?
[Coordinating FF values across environments] for reporting needs time and a lot of co-ordination.
"...the major issue we have is finding time to remove flags when we have product roadmaps to keep up with. Our PO is going to allocate time on a per sprint basis starting next year to remove them."
...removal of code [around feature flags] needs more analysis on risk and testing work flows
why is cleanup important?
Technical debt Performance risk
Risk of errors Developer productivity
Over time, our system accumulates outdated LaunchDarkly feature flags that clutter the codebase, increase maintenance overhead, and pose risks to stability.
PROBLEM STATEMENT
proposed Solution
We propose an AI-powered solution to automatically identify and remove obsolete flags, improving efficiency and reducing technical debt.
AI-Driven Development Backend was designed and built using AI tools like Windsurf and GPT-4, accelerating prototyping and reducing manual effort.
Smart Orchestration Layer Java Spring Boot backend integrates with LaunchDarkly to serve retirement-ready flags and trigger intelligent automation.
Backend Approach
Seamless Integration - Backend applies AI-generated changes, commits them, and creates ready-to-review pull requests with detailed modification context.
Automated Refactoring GPT-4 performs deep code analysis and generates production-ready changes to safely remove feature flags across Java, JSP, and JS.
demo
Integration
Intelligent Flag Identification
Risk Aware Recommendations
Automated suggestions integrated to propose removals
Conclusions
Time and Effort Reduction
Cross-Team Visibility
Thank you!
'A clean codebase is like a crisp autumn morning—refreshing and productive.'
- COPILOT
Prompt Engineers
Niya anil Yashashree Majalikar Ganesh Ravendranathan
Saba Saba Saiprathap sankineni kishan trivedi
TECH STACK
Spring Boot
Angular
Azure AI Foundry
Windsurf
Demo screenshots