Java offers advantages over Python in enterprise settings due to performance, dependency management, and its mature ecosystem for large corporations.
LangChain4j simplifies AI development in Java by providing abstractions for models, memory, prompt templates, and function calling via AI services.
A practical demonstration shows how to build a theme park chatbot that answers questions using document retrieval and function calls.
Simple agent architectures that send all context to the model at once lead to higher costs, slower responses, and increased hallucinations.
LangGraph4j provides a framework for defining complex, multi-step agentic workflows as stateful graphs with nodes, edges, and a shared state.
A code walkthrough shows how to define a graph's state, create nodes for functions, and use conditional edges to route the execution flow.
LangGraph4j supports pausing a graph's execution at a checkpoint, allowing for human review or input before resuming the process.
A multi-step graph demonstrates an AI agent that categorizes emails, searches the web, drafts responses, and uses guardrails to verify its own output.
A QR code is provided to download the presentation slides, which contain links to all the demonstration code shown in the talk.
30:01Mario Fusco
29:18Daniel Oh & Kevin Dubois
31:59Juarez Junior
26:42Julián Duque
24:22Philipp Schmid
29:03Jörg Neumann
32:02Kevin Dubois
31:53Timo Salm
Jobs that call for the skills explored in this talk.
Deutsches Medizinrechenzentrum GmbH
4Com GmbH & Co. KG
CGI Group Inc.
Köln, Germany
auteega GmbH