Conferences

Drawing on the measured facts behind a real 11-day sprint, we'll walk through a practical playbook for building with AI agents:

• The mindset shift: from writing every line to piloting a fleet of agents — and an honest with-vs-without-AI comparison
• Write the specifications so agents know exactly what to build
• Build the test harness that lets agents verify their own work
• Split the work and run agents in parallel, each on its own branch
• Let a few deep plans run overnight, then review and merge over coffee

Demo-driven and language-agnostic: expect concrete, opinionated advice drawn from really shipping with parallel AI agents.

In this session, we'll cover:

• 58 panels in 9 groups, served from your own application at /bootui — no agent, no sidecar, nothing to configure
• Advisors: 437 rules written with AI across architecture, security, Hibernate, memory and dependencies, each naming the cause and the fix
• Reading the live application: requests, SQL traces with N+1 detection, transaction boundaries, exceptions with regression detection, and the JVM
• The MCP server: the same data exposed to your coding agent, so it stops guessing and reads the running app
• One codebase, three stacks: Spring MVC, Spring WebFlux and Quarkus — and what it takes to stay honest when a capability doesn't exist on all of them

A demo-driven session on a tool built for the inner loop: local-only, fail-closed, and designed to explain rather than dump.

In this session, we'll cover:

• The rapid evolution of AI agents: from completion to autonomy, with agent mode, plan mode, and parallel agents using Git worktrees
• The Model Context Protocol (MCP): how agents connect to databases, APIs, and external tools in real time
• Agent Skills: how to specialize an agent for a specific domain — with a live demo of dr-jskill, generating a full Spring Boot 4 + Java 25 + Vue.js application
• IDE integration: GitHub Copilot in VS Code and IntelliJ IDEA, plus the new GitHub Copilot desktop app for running multiple agentic sessions in parallel
• Best practices for Java: AGENTS.md, JDTLS (Eclipse JDT Language Server), and why the harness around the model matters more than the model itself

Expect plenty of live demos and concrete, opinionated advice drawn from real-world experience building and shipping with AI agents.

Drawing on a month of real GitHub activity — hundreds of merged pull requests and billions of tokens across 11 models — we'll walk through a concrete 24-hour playbook:

• The mindset shift: from developer to manager of agents — you don't type faster, you ship wider
• Prep in the evening: write the house rules in AGENTS.md, give every agent a build + test harness, and set clear guardrails
• Build overnight: run several agents in parallel, each looping on its own branch / worktree — build, test, open a PR, pick up the next task
• Review over coffee: triage tens of PRs, merge the wins, drop or re-task the rest, then relaunch for the day
• Guardrails that keep it safe: tests before trust, small reviewable PRs, the right model for each task, and reusable skills

Language-agnostic and demo-driven: the playbook applies whatever your stack. Expect concrete, opinionated advice from really shipping with parallel AI agents.

No prior Spring Boot experience is required — just a terminal and an editor. Over ten chapters, each ending with a checkpoint, you will:

• Install the toolchain: Node.js, Java, Docker, GitHub Copilot CLI, and the dr-jskill Agent Skill
• Generate a full Todo application from a single prompt: REST API, Hibernate, PostgreSQL, Vue.js front-end, and Docker for local development
• Explore the anatomy of the generated code, then extend it with a user system and per-user filtering
• Polish the front-end, add unit and integration tests, and tune performance
• Deploy it: run in Docker, build a native image, and outline a cloud deployment

You leave knowing how to drive an AI coding agent to produce real, production-shaped Spring Boot applications — and how to understand and extend the code it writes.