AI | Matheus Oliveira

A slip-box of notes wired into an AI agent

Using Zettelkasten with Agents

How structured note-taking with Zettelkasten became the memory layer for my AI agents, and how it saves me tokens, time and money.

Minimalist diagram showing a central 'Agent' node connected to four modules: Soul, Memory, Skills and Tools, on a dark purple background

Qlawkus: Build Personal AI Agents in Java with Quarkus

Introducing Qlawkus, a set of Quarkus extensions for building personal AI agents in Java: memory that’s automatically injected, procedural skills, voice-enabled messaging, and composition via agent.yml.

SNES 16-bit pixel art of a robot wizard in a red cape holding floating weather and geocoding tools, with code runes in the background

AI Agent Tools in Quarkus LangChain4j: When the REST Client Becomes the LLM's Right Hand

Build an AI agent that calls real REST APIs as tools, without writing any orchestration logic. REST Client + @Tool = automatic tool. The LLM decides when to call, which to call, and with what arguments. Everything runs locally with Ollama via Dev Services, zero API key.

Java AI Libraries Compared: LangChain4j, Spring AI, and Quarkus LangChain4j

I built the same AI app with LangChain4j, Spring AI, and Quarkus LangChain4j. One of them took 92 lines. The others took 192 and 180.