Introduction: The Challenge of Building Agentic Systems When I started building agentic applications, I encountered a fundamental challenge: the landscape of agentic frameworks and libraries is evolving at breakneck speed. Each new project brought new tools - LangChain one month, Google ADK the next, then Agents SDK, and so on. Each framework had its own patterns, its own organizational philosophy, and its own way of thinking about agents. This constant churn made it difficult to maintain consistency across projects or transfer knowledge between them. ...

Building a Production-Ready Agent Stack: Part 4 - Agent Integration & Streaming In Parts 1-3, we built the foundation: Docker containers, FastAPI backend, database with migrations, and Auth0 authentication. But we still don’t have the core feature: intelligent agents that can remember conversations, call tools, and stream responses in real-time. Today, that changes. ...

Building a Production-Ready Agent Stack: Part 3 - Authentication & Security In Parts 1 and 2, we built a solid foundation: Docker containers, FastAPI backend, database with migrations, and clean architecture with repositories and services. But right now, anyone can access anyone’s sessions and messages. Time to fix that. ...

Building a Production-Ready Agent Stack: Part 2 - Backend Core & Database Our FastAPI server is running, but it’s just a health check endpoint talking to no one. Time to build the persistence layer - database models, repositories, domain services, and the REST API that will power our agent conversations. ...

Building a Production-Ready Agent Stack: Part 1 - The Foundation Welcome to the first post in a series where we build a real, production-ready AI agent application from scratch. No shortcuts, no toy examples - just the patterns you’d actually use in production. ...