Watch a conference talk from GOTO Copenhagen 2024 where Software Architect Thomas Vitale demonstrates how to build production-ready LLM applications using Java and AI through the practical example of creating a composer assistant. Learn about core architectural patterns for integrating AI capabilities into existing software systems, including templated prompts, agent tools, and external knowledge integration. Explore the Java ecosystem's AI capabilities through hands-on demonstrations covering text classification, semantic search, question answering with documentation, structured data extraction, and speech transcription. Discover practical solutions for production challenges like local development with language models, token-based observability, and ensuring resilience across AI-orchestrated integrations. The 34-minute presentation culminates in a live demonstration where Vitale composes and performs music for an audience-chosen movie scene using the AI-assisted application.
Overview
Syllabus
Intro
GenAI
The WHY factor
Demo
Text classification
Demo
Semantic search
Demo
Question answering with docs
Demo
Structured data extraction
Speech transcription
Demo
Final product
Outro
Taught by
GOTO Conferences