The Era of AAP: Ai Augmented Programming using Modern Java
summary
Claims Java can run end to end for AI-augmented programming: a Java IntelliJ plugin Devox Genie wires local LLMs (llama.cpp, Llama 3.2/3.3, Mistral) and cloud models (Anthropic Sonnet, Google Gemini) and mixes them to review each other; demo sends a 140k‑token project and gets a multi‑model review (32s, ~168k tokens). Shows Java inference projects: llama3‑java single‑file (≈2,545 lines, Q4/Q8, float16) and J‑Lama with REST, claims native Java can beat llama.cpp and supports local, offline, privacy workflows.
AI-written summary · about
chapters (12)
- 0:00Welcome to AI Augmented Programming
- 1:52Local vs cloud LLM providers
- 4:00Privacy with local models
- 8:15System prompts and chat memory
- 12:06Java LLM tools: LangChain4J and Spring AI
- 13:30Devox Genie Java plugin
- 24:51Java for LLM inference
- 26:41JLlama and Llama3.java engines
- 37:14LLM sharding strategies
- 39:01Future LLM optimization
- 41:50AI agents for Java devs
- 45:05Q&A
related talks
38mThe Era of AAP: Ai Augmented Programming using only Java
46mThe Era of AAP: Ai Augmented Programming using only Java
46mThe Era of AAP: Ai Augmented Programming using only Java
58mGenesis Keynote
51mPractical LLM Inference in Modern Java
30mDevoxxGenie: Your AI Assistant for IDEA
49mTame your Llama, run AI in Java
47mPractical LLM Inference in Modern Java