The State of Large-Scale Code Modernization
summary
A demo shows a desktop agent called Motti that runs across business units, ingests a lossless semantic tree (compiler metadata and deps) and uses a recipe marketplace to determine Java versions (finds Java 8, 11, 17 across repos). The system exposes ~4,000 tools by inserting a recipe search gateway to bypass a ~100-tool API limit, then executes recipes, returns tabular results and issues SQL reductions. Integrations include Diffblue test synthesis, Azul runtime reachability to remove dead code, and Volcheck vulnerability data; the stack supports deterministic, agentic modernization and options for backpatch versus forward upgrades.
AI-written summary · about
chapters (7)
- 0:00Welcome and event overview
- 3:57Demo of Motti agent system
- 10:20Tool function calling trends
- 14:58Integrations with Diffblue and Azul
- 19:38Vulnerable dependencies and prioritization
- 21:48Technology lifecycle management debate
- 27:05Closing remarks and Q&A
related talks
59mTools & Tips to Modernize Your Java code with AI
42mClosing the Agentic DevOps Loop with Microsoft and Moderne
27mNavigating Application Modernisation - Leveraging Gen-AI
40mHow BNY Uses Moderne to Make AI Agents Cheaper and More Accurate
41mHow Agentic Coding Can Help You Migrate Java Applications Faster
20mBuilding intelligent, secure, and scalable autonomous systems
41mIncremental Modernization through CodeConcise
28mApplication modernisation with generative AI