MAX Research Collective

Published · Series · Reference

The Agentic AI Builder’s Playbook

A 27-part series that maps production-grade agentic AI architecture end to end — written from inside the build, not the hype cycle. It was published as 25 parts; parts 26 and 27 extend it.

Start with part one Agentic AI writing
WritingAgentic AI27 Parts

What it covers

  • The nine-layer agent stack and the twelve core components of an agentic system.
  • The MCP and A2A protocols every engineer now has to know.
  • The RAG family — classic, GraphRAG, agentic, and vectorless retrieval.
  • Agent memory and context engineering.
  • Multi-agent choreography versus orchestration, and the patterns that decide whether a system holds together under load.

How it reads

Each part stands on its own as a field note with the diagram, the trade-off and the decision framework — when to reach for an agent, and just as often, when not to. Each part ends with a link to the next.

All 27 parts

  1. The 9-Layer Agentic AI Stack: How Modern Organizations Build Run and Secure AI Agents at ScalePart 1
  2. MCP vs A2A: The Two Protocols Every AI Engineer Must Know in 2026Part 2
  3. AI Agent Memory: Why Short-Term Memory Is Killing Your AgentsPart 3
  4. One Architecture Diagram That Explains Every AI AgentPart 4
  5. Classic RAG vs GraphRAG vs Agentic RAG: Choosing the Right Retrieval PatternPart 5
  6. When NOT to Use an AI Agent: A Practitioners Decision FrameworkPart 6
  7. The 2026 AI Engineer Is NOT What You ThinkPart 7
  8. What Happens When You Call Any LLM API: The Full Journey in 400msPart 8
  9. Claude Code vs Codex CLI vs Gemini CLI: The Agentic CLI WarsPart 9
  10. Your AI Strategy Has a Single Point of Failure: Your DataPart 10
  11. The 12 Core Components of an Agentic AI SystemPart 11
  12. Claude Code 3-Layer Memory Architecture: What 512K Lines of TypeScript RevealedPart 12
  13. How to Design a CLAUDE.md That Actually WorksPart 13
  14. Claude Code Is Not a Folder Structure: Its an Agentic Operating SystemPart 14
  15. Dashboard BI vs Agentic Analytics: The Next Wave of Business IntelligencePart 15
  16. The Zero Dollar AI Stack: Building a Full AI App Without Paying a CentPart 16
  17. Your AI Agent Just Broke 3 Systems in Production: Heres WhyPart 17
  18. Understanding Agent Skills Hooks Subagents and Plugins: The Complete ArchitecturePart 18
  19. The SKILL.md Universal Standard: How AI Agents Share Reusable ExpertisePart 19
  20. The Open-Source AI War Just Exploded: April 2026 ScorecardPart 20
  21. Claude Code Agent Teams: Production Patterns for Multi-Agent CollaborationPart 21
  22. The 6 Processors Powering Modern AI: CPU GPU TPU NPU LPU and DPU ExplainedPart 22
  23. Vectorless RAG vs Traditional RAG: Navigating by Structure Not SimilarityPart 23
  24. Repository Intelligence: The Skill Replacing Prompt Engineering in 2026Part 24
  25. The AI Code Review Problem: Why the Same AI That Writes Your Code Cant Review ItPart 25
  26. The Pre-Existing Stack You Can’t IgnorePart 26
  27. Choreography or Orchestration: Picking the Spine of Your Multi-Agent SystemPart 27

Follow the collective

New research, projects and field notes as they publish. Following happens on LinkedIn — we never see an address, and there is nothing to unsubscribe from.

Or get it by email

An occasional summary of new research, each with a link to the full piece.