← all paths traced services → multi-agent system

Agentic AI on the JVM

Centralize error handling behind one contract, trace requests across services and LLM calls with OpenTelemetry, then build a multi-agent investigation system with LangChain4j, Neo4j, and Qdrant.

3 steps · 40 min total
  1. 01
    note Backend

    Centralized exception handling in REST APIs: the same handler contract in Spring and FastAPI

    REST API exceptions fail in two ways: swallowed silently, so the error becomes invisible, or leaked raw, exposing internals to the client. A centralized handler with a structured error contract fixes both, implemented the same way in Spring and FastAPI.

    May 03, 2026 · 7 min read
  2. 02
    essay Infrastructure

    Observability in a hybrid microservices stack: Spring Boot, FastAPI, and LangChain on Kubernetes

    In distributed systems, a single request can cross multiple services and runtimes. When something breaks, we need to easily reconstruct the flow of that request through logs, metrics, and traces, instead of manually navigating unrelated application logs with no correlation.

    May 05, 2026 · 13 min read
  3. 03
    essay AI

    InvestigatorAI: a near-production multi-agent AI system

    InvestigatorAI answers investigative questions that a single retrieval step cannot solve: a supervisor agent decomposes the query and dispatches five specialist agents over a corporate ownership graph in Neo4j and a document collection in Qdrant, with Langfuse tracing every step. Built with LangChain4j on Spring Boot.

    Jun 03, 2026 · 20 min read
  4. path complete