Architecture

21 October 2026·18 min·Advanced·Published
Decide, don't generate
Most enterprise model calls are decisions, and decisions have better tools than generation
A large share of production model traffic is routing, triage, classification and validation — small decisions paid at the price and latency of a general-purpose chat model. Decision models make the output typed and the probability calibrated. This paper looks at where that trade is worth making, what the vendor benchmarks do and do not show, and what has to be true architecturally before a decision layer is safer than a prompt.
#AI#Architecture
• #decision-models
• #system-one
• #llm-routing
• #calibration
• #agents
21 October 2026·18 min·Intermediate·Published
A serious production on a single machine
One entry point, a disciplined edge, and a trade that should be stated before it is discovered
A single machine can host a real production stack, provided the edge is treated as an architecture rather than a pile of containers. This paper describes the shape in the abstract: exactly one public entry point, a filtering layer behind it, application servers that never see the internet, a management plane that is not reachable from a public interface, and an update ritual that can be rehearsed. It names no products on purpose — the principle is the transferable part, and the configuration is the part worth keeping quiet. It ends with the limits of a one-machine trade, and with what changes when the system outgrows it and the edge moves onto dedicated equipment.
#Cloud#Architecture#DevSecOps
• #vps
• #edge
• #waf
• #operations
• #architecture

No ideas under this topic yet.

No research notes under this topic yet.