Executive Briefing: Point an agent at your calendar and your repo, and it will show you the rules your company is actually running. Here are the 15 I wrote for mine.

Source: Nate’s Newsletter Date: 2026-07-12 URL: https://natesnewsletter.substack.com/p/ai-native-company-rules

Summary

An essay arguing that companies run on unwritten rules shaped by past resource scarcity, and that making an organization legible to AI agents requires rewriting those rules as explicit, enforceable constraints. The author (paywalled past the framework) offers a methodology — separating values/rules/runtime-checks/human-appeals, a five-question worksheet for turning informal norms into codified rules, and a five-rung enforcement ladder from “value” to “hard block” to “human-owned decision” — illustrated with one rule implemented as actual automation (calendar check, Slack challenge, deadline, then escalation).

Implications

Feeds the agentic engineering thread, specifically the governance-of-agent-authority sub-thread that’s been recurring in Claude Code’s own releases (consent provenance, permission scoping, auto-mode gating). The framing is notable in reverse: instead of a vendor hardening what an agent is allowed to do, this is an operator arguing that organizations need to externalize their implicit rules into machine-checkable form before agents can be safely pointed at company processes. Low-to-medium signal as a primary source (single newsletter essay, methodology gated behind paywall), but a useful data point on how operators outside the model/tooling vendors are thinking about agent-company interfaces.