AI-Powered Form Wizards: Chat, Click, Done
summary
Prototype shows an AI form helper that converts chat into validated form data, using LangChain4j and Quarkus with a pgvector-backed RAG pipeline, Java bean validation for constraints, and small Java tools to de-hallucinate LLM outputs. Demo pre-fills tax-like forms (13 tabs) and catches a missed checkbox that cost >€1,000, parses images and PDFs to extract fields, runs chained and parallel agents, and ships code on GitHub with OpenAI or Olama switches for local models.
AI-written summary · about
chapters (7)
- 0:00Welcome and introduction
- 6:03Pain points of traditional forms
- 13:08Building AI agents step by step
- 20:57Live demo: Puppy adoption form
- 27:02True agents and multimodality
- 33:49Security and future of AI
- 36:40Q&A and closing remarks
related talks
51mAI-Powered Development: Hands-On Techniques for Immediate Impact
47mCreating AI-infused Apps, the Easy Way 🤙
41mFrom Chat to RAG to MCP: Enhancing Java Applications with AI
01h 01mCreate #AI Infused #Java Apps with #LangChain4j
45mFrom Click to Conversation: Let's Replace Buttons and Forms with an LLM!
43mAI w/o Spaghetti: Clean Architecture in the Age of AI
41mAI without spaghetti: Clean architecture in the age of AI
50mDeveloping AI applications in Java with Quarkus and Langchain4j