"Sudo, Make Me a Sandwich!" – Live Coding With Agentic AI in IntelliJ IDEA
summary
Live demo uses JetBrains Junie inside IntelliJ IDEA to generate a plan, create a service layer from controllers, add logging and utility classes, run tests and iterate; agent runs under a Brave allowlist to execute test commands. Workflow compares VIP coding (vague “improve it” prompts) with directive coding (explicit checklist in docs), shows Junie creates files and refactorings and fixes some failures but leaves a few tests red after its iterations, so outcome is partial success requiring human guidance.
AI-written summary · about
chapters (19)
- 0:00Welcome
- 1:03AI programming levels
- 5:43Vibe vs directive coding
- 7:12Spring Pet Clinic setup
- 10:04Starting vibe coding
- 13:00Directive coding intro
- 17:34Improvement plans with Juni
- 21:30Analyzing architecture plan
- 25:47Creating tasks
- 28:40Implementing plan
- 36:17Test failures handling
- 41:36Agent limits and retries
- 43:52Better agent guidelines
- 47:09Data privacy choices
- 50:01IntelliJ AI plugins
- 55:17Agents vs manual coding
- 57:25Managing expectations
- 59:12Reviewing changes
- 1:00:15Q&A
related talks
08mAI inside the IDE
58mAI Assistant: Don’t Code Alone
41mBuilding with AI in Kotlin
42mDeveloper Productivity With IntelliJ IDEA
43mBe more productive with IntelliJ IDEA
41mAI Tooling for Scala Developers - What JetBrains Brings to the Table
35mGoing Beyond Copilot with AI Agents
27mIntelliJ IDEA Tips & Tricks