Java + LLMs (Build LLM Apps in Java with JakartaEE)
summary
Argues Java is the practical choice for building LLM apps (not training), using LangChain for Java with Jakarta EE and a WebSocket UI. Shows a streamed chatbot, session memory persisted to Postgres, function-calling tools that fetch webpages, compile and run generated Java, and a generator that builds full projects. Demonstrates RAG: documents embedded into pgvector, splitting into five-character chunks changes retrieval and causes false attributions. Notes GPT‑3 scale (~300 billion tokens) and that outputs remain unreliable without careful prompts and chunking.
AI-written summary · about
chapters (8)
- 0:00Introduction and setup
- 12:39Basic chatbot with Jakarta EE
- 20:17Prompt engineering and AI services
- 32:03Memory and context in conversations
- 37:04Function calling and tools
- 48:48Project generation with AI
- 49:29Retrieval augmented generation demo
- 59:41Q&A and conclusion
related talks
02h 49mJava meets AI: Build LLM-Powered Apps with LangChain4j
57mPure Java Enterprise AI/LLM Integration (EAI 2.0)
53mJava meets AI: Build LLM-Powered Apps with LangChain4j
52mLivecoding: Creating Beautiful Java Code With LLM and Agents
47mJava Meets AI: Integrating LLMs into Your Applications with LangChain4j
46mGenerative AI and Java
45mBuilding Intelligent Enterprises: AI-Powered Java with LangChain4j & CDI
34mConcerto for Java & AI – Building Production-Ready LLM Applications