AI assisted legacy modernisation : From UX to code
summary
Bahmni’s decade‑old EMR (used in ~50 countries, ~1M LOC) migrates UI and data model using Generative AI for UX and coding agents for code, targeting FHIR compliance. Workflow uses AI design reviewer and persona chat (QoQ) to cut a display‑control redesign from ~8 days to 3, retrieves designs into Figma and prototypes with Bolt, and uses a coding agent in VS Code (Anthropic Sonnet 3.5, mounted docs and repo) to analyze and rebuild 30 widgets; consultation module ships in ~3 months and developer steps run ~96x faster.
AI-written summary · about
chapters (12)
- 0:00The Bahmni challenge
- 2:13Legacy UX limitations
- 3:33Modernization strategy
- 4:35Migration hurdles
- 6:34Defining the workload
- 8:06AI-assisted UX journey
- 13:28Impact of AI on design
- 18:55Lessons in AI design
- 20:56AI-assisted development
- 26:21Developer productivity gain
- 28:11Critical technical warnings
- 30:14Key takeaways
related talks
40mHuman Centric Software Development Augmented by AI
43mAI accelerated legacy modernisation
16mAI-Powered Mainframe Modernization
17mAI-Driven Frontend Development: Beyond Traditional
33mAI assisted software delivery
55mAI Assisted Engineering
27mSoftware engineering with GenAI
28mApplication modernisation with generative AI