Intel / post

Imagine if naked people were stupider. It turns out, naked models actually are.

Source: x.com
Author: Garry TanTop Viber
Date: Indexed May 4, 2026

Context

Garry Tan argues that AI reliability should be assessed across the complete application: the model, its instructions, its tools, and the checks on its output. His essay proposes assigning judgment to the model while giving precise operations, such as retrieving records or performing calculations, to ordinary software.

The connecting mechanism is task routing. A reusable procedure tells the agent which tool to use, and the surrounding program executes it and returns the result. This creates individual steps that developers can test. Tan acknowledges that procedures, tools, and verification can themselves be wrong; his proposal makes failures more inspectable, rather than establishing that the whole system is trustworthy.

Key takeaways · AI-distilled

The sharpest evidence is Anthropic's own: the leaked Claude Code source is roughly 512,000 lines of scaffolding — repo , , purpose-built tools, session memory, sub-agents — around a model its makers won't ship bare.
The vocabulary is worth stealing. A file is markdown teaching a procedure, not a prompt. A resolver is a routing table picking which skill to read. Deterministic code handles what must be exact. The is the thin loop.
The author concedes the cost: 512,000 lines is not simple, harnesses fail, skills encode wrong procedures, and a verification layer is only as good as the invariants it checks. The claim is tractability, not ease.

Terms in this piece · Glossary

context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
prompt caching — Reusing the model's processed form of a repeated prompt prefix so subsequent calls skip re-reading it, cutting cost and latency substantially.
agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.
agent harness — The scaffolding around a model that turns it into a working agent — the loop, the tools it can call, and the rules for when to stop.

Key quotes

“He's testing the engine on a bench and concluding that cars are unsafe.”

“Because harnesses make the system trustworthy, even when the model inside it is not.”

“That's 512,000 lines of evidence that model reliability is an engineering problem, not a philosophical one.”

“The claim is that they turn model unreliability from a reason-to-stop into a problem-to-solve.”

Read the source x.com