AI works great in a demo, then falls apart on your real codebase. In this episode, three practitioners show you how to modernize large legacy systems with AI, get AI agents to actually ship working code, and get AI writing code that passes your own review. Watch the demos and real code walkthroughs, with an honest look at what still breaks.
Watch three practitioners modernize legacy systems, run coding agents, and get AI-written code through review.
On demand
Leave your email to get the Episode 8 recording and materials. Watch all three talks at your own pace.
What are the talks about
Two years of modernizing large, business-critical systems with AI across real teams, not a greenfield demo. Simon shares what worked at scale, what quietly wasted time, and where the approach still hit a wall in production.
Rival coding agents, one shared context, and independent review gates shipped a real MVP in three days without a human at the keyboard. Baruch opens the rules, review layers, and failures behind the software factory.
Five ways of working with AI, four dead ends, and one process that produced code worth shipping. Pasha shows the prompts, artifacts, and decision-making workflow he now uses in real Java development every day.
What you will learn
Episode 8 is about the AI questions Java teams actually hit in production: what to build on, what to hand off, what to keep human, and where AI still cannot be trusted.
Get a model for AI-driven modernization at scale. Learn how to write specs an agent can actually use, how much generated code is worth reviewing, and where the approach still breaks, including projects that got thrown away.
See how rival coding agents share one context and check each other through independent review gates, so they ship working software instead of spinning in endless bot loops. You'll leave with a practical pattern for building your own lean setup.
Learn the workflow that keeps your standards in AI-generated code. You make the decisions and AI does the typing, so you stop shipping code you would reject and stop losing time fixing it.
Headliners
Java Champion · Oracle ACE Pro · Creator of the AI Unified Process
Simon has three decades as an architect, consultant, and trainer, and has spent the last two years modernizing large business applications with AI across multiple teams and legacy systems. In his talk, he shows what actually works at scale, what quietly wastes time, and honestly where AI still cannot do the job.Simon is an architect, consultant, and trainer with three decades of experience. Over the past two years, he has modernized large business applications with AI across multiple teams and legacy systems. At JRush, he explains what works at scale, what wastes time, and where AI still cannot do the job.
AI modernization, legacy systems, spec-driven development
Head of Developer Relations, Port.io · Java Champion · Co-author, “Liquid Software” & “DevOps Tools for Java Developers”
Baruch did Java before generics, DevOps before Docker, and DevRel before it had a name — and led DevRel at JFrog from ten people to a $6B IPO. Obsessed with how AI agents actually write code, he built a “software factory” of rival AI agents that shipped a real MVP with no human at the keyboard. His session opens the coding-policy plugin, the review gates, and the failures behind it — a practical pattern you can build yourself.Baruch built a “software factory” of rival AI agents that shipped a real MVP with no human at the keyboard. At JRush, he opens the coding-policy plugin, review gates, and failures behind the setup.
AI agents, context engineering, agentic workflows
Developer Advocate, BellSoft
Pasha is relentless about code quality. After burning through four AI workflows that did not hold, he found the one that does, where AI writes but the decisions stay his, and nothing ships below his bar. His talk puts all five on screen, with the artifacts from real work.
AI-assisted coding, code review, developer workflow
Free Java web conference
JRush is a free BellSoft series with expert talks, recordings, and technical materials for teams building, running, and securing Java applications. Episode 8 is about putting AI to work in real Java systems, from rival coding agents shipping real software, to modernizing legacy at scale, to writing code you would actually ship.
Free Java web conference
FAQ
Leave your email to get the Episode 8 recording and materials in your inbox.
JRush is a free online series about Java development, cloud-native engineering, and the practical tools and workflows modern Java teams use.
Yes. The recording and materials are free. Submit the form to receive them by email and watch at your own pace.
Java and Spring developers, DevOps and platform engineers, architects, and technical leads working with or deciding on AI in real Java systems, from greenfield to legacy.
You will receive the Episode 8 recording and materials covering AI-driven legacy modernization, rival coding agents, and AI-assisted Java development.
Watch on demand
Catch up on JRush Episode 8. Three practical talks on AI agents, legacy modernization, and code you can trust. Get the free recording and materials in your inbox.
We will send the Episode 8 recording and materials to your email.