Skip to content
AI Anytime
← Paths

02

Production Systems

The gap between a demo and a system people depend on is evaluation, memory, and observability.

6 lessons · ~60 hours

0 / 6
  1. 01

    LangGraph

    Graph orchestration with real state. Where it beats a plain loop.

    Agentic RAG using LangGraphPerplexity Lite
    Build Your own Claude Code with LangGraph and MCP
    Microsoft’s Agent Governance Toolkit Explained + LangGraph Demo
  2. 02

    CrewAI and role-based agents

    Multi-agent collaboration, plus the coordination cost nobody mentions.

    Agentic RAG using Crew AICrewAI AgentOps
    Which AI Agent Framework Wins? LangGraph vs CrewAI vs AutoGen vs OpenAI Agents SDK
    Build a Lead Generation AI tool with Multi AI Agents 🤖
  3. 03

    Model Context Protocol

    Build an MCP server, connect a client, understand the boundary.

    Awesome MCP ServerAI Agent with MCP Tools
    4 - Windsurf + Supabase MCP (Model Context Protocol)
    MCP (Model Context Protocol) Workshop
  4. 04

    Evaluation

    Harnesses for retrieval quality and end-to-end task success.

    RAG evaluatorEvaluation of LLMs and RAGsSelf Healing RAG
    Testing Framework Giskard for LLM and RAG Evaluation (Bias, Hallucination, and More)
    Evaluate LLMs with Language Model Evaluation Harness
  5. 05

    Memory

    What to persist, what to summarize, what to throw away.

    Agent Memory PlaygroundMem0 memory layer
    Mem0 + Neo4j | Memory Layer for Agentic AI System
    19 - OpenMemory MCP: Secure and Local Memory for AI Agents
  6. 06

    Observability

    Tracing, cost attribution, and making non-determinism debuggable.

    agent watchAgentOps tracing
    AI Community Townhall: Giveaways & AI AgentOps Library Announcements
    Monitor Your AI Agents with AgentOps: Build Production Grade AI Agents

Next

Scale and Infrastructure

→