The Era of AAP: Ai Augmented Programming using only Java

on YouTube ↗46m

" summary

Shows a full-Java inference stack that runs LLMs without Python, with a single-file llama3.java implementation and a Swing IntelliJ plugin Devoxx Genie (≈9,000 downloads) that chains local and cloud models. Demonstrates local Llama 3.1/Qwen and cloud Claude and Gemini, cites Claude ~200k-token and Gemini 2M-token contexts, and claims native-image preloading drops time-to-first-token to about 1 ms while typical project scans hit ~140k tokens; privacy and on‑prem sharding options are explored.

AI-written summary · about

" chapters (10)

1. 0:00AI-augmented programming intro
2. 1:07Local vs cloud LLM providers
3. 8:36AI will not replace jobs but developers using AI will
4. 11:05Devoxx Genie IntelliJ plugin
5. 18:31Pure Java inference with GraalVM
6. 27:17llama3.java implementation details
7. 33:34Demo: llama3.java locally
8. 39:56Integrating llama3.java
9. 41:43LLM sharding for large models
10. 44:25Future of agentic AI in Java

" related talks

• 38mThe Era of AAP: Ai Augmented Programming using only JavaStephan Janssen · Voxxed Days Thessaloniki 2024
• 46mThe Era of AAP: Ai Augmented Programming using Modern JavaStephan Janssen · Voxxed Days Ticino 2025
• 46mThe Era of AAP: Ai Augmented Programming using only JavaStephan Janssen · Voxxed Days Bucharest 2025
• 58mGenesis KeynoteStephan Janssen · Devoxx Belgium 2024
• 47mPractical LLM Inference in Modern JavaAlina Yurenko, Alfonso² Peterssen · Voxxed Days Zürich 2025
• 57mPure Java Enterprise AI/LLM Integration (EAI 2.0)Adam Bien · Devoxx Belgium 2024
• 51mPractical LLM Inference in Modern JavaAlina Yurenko, Alfonso² Peterssen · Devoxx Belgium 2024
• 30mDevoxxGenie: Your AI Assistant for IDEAGunter Rotsaert · Devoxx Belgium 2024