Developer proposes a 7-layer OSI-style model for AI agents, betting most failures live in planning and feedback
Klutzy-Trust7439 · reddit · 2026-11-22
A developer on Reddit shares a 7-layer framework for thinking about AI agents, modeled loosely on the OSI stack:
- Model — predicts next tokens, can do nothing alone
- Memory — history, notes, retrieved docs
- Tools — search, browser, code execution, APIs
- Planning — decomposing goals, choosing tools
- Actions — clicking, typing, writing files
- Guardrails — limits, spend caps, human-in-the-loop, logs
- Feedback — reading own output and retrying
A chatbot is layer 1 plus a textbox; layers 2-3 make an assistant; 4-7 make an agent. His bet: most real-world failures happen in layers 4 and 7 — plans look fine, one step silently fails, and the run "succeeds" with a bad result. He also wonders whether identity/permissions deserve their own layer, and plans to build a minimal agent from scratch to see which layer breaks first.
More from coding & agent
- A2A protocol walkthrough: how triage and billing agents collaborate on a refund flow — adnan_hashmi · 2026-11-23
- Dev Migrates From Vercel to Cloudflare in 35 Minutes With Claude, Cutting Bill From $70 to $10 — aloncarmel · 2026-11-23
- How A2A protocol works for agent-to-agent delegation in Microsoft Foundry — adnan_hashmi · 2026-11-23
- Aiden Bai demos new dev tool that runs fully locally in the browser — aidenybai · 2026-11-23
- Lantern: an open-source free-running agent sandbox for Qwen 27B on Codex CLI — MikePFrank · 2026-11-23
- Only-Knowing-Code Developers at Risk as the Agentic Software Factory Reshapes Engineering — Pavan_Belagatti · 2026-11-23