An agent stack diagram says production AI is 90% architecture, not prompts
theomitsa · x · 2026-07-23
A shared diagram argues that production AI is mostly systems engineering, not just prompt writing or model choice.
It contrasts a toy pipeline:
- Prompt → Model → Tool → Response
with a production stack built from:
- Model
- Context
- Memory
- Tools
- Skills
- Orchestration
- Identity
- Policy & guardrails
- Observability
- Evaluation
- Runtime
The central thesis is that the model may be only 10% of the agent architecture, while the remaining 90% decides whether the system survives in production. The chart also lists what breaks when each layer is missing, plus best practices such as using the smallest reliable model, making plans observable, automating evals, and designing for graceful degradation.
Related event: AI Agent Engineering Shifts to Layers and Loops(5 posts)→
More from coding & agent
- huggingface_hub 2.3.0 cuts 1M-file listings to 20s, adds live Bucket following — lhoestq · 2026-10-27
- Nous Research ships Advanced Delegation plugin letting Hermes Agent route subagents to different models — NousResearch · 2026-10-27
- A dentist with a 3D printer discovered OpenAI Codex — and never looked back — gabrielchua · 2026-10-27
- Stanford's PaperFormalizer turns research papers into Lean-verified versions with author in the loop — StanfordUniversity · 2026-10-27
- Hugging Face TRL makes score centering + masked importance sampling default in async GRPO — _lewtun · 2026-10-27
- a16z: agents follow tech's oldest law — chaos first, then rein it in — a16z Newsletter · 2026-10-27