Why Clay Has an AI Writing Policy

Clay’s writing policy has one blunt premise: whoever hits publish still owns every sentence, no matter how it got drafted. Engineering leader Sophie Alpert explained the policy after AI-generated internal documents at Clay started showing up bloated, hard to read, and full of the same generic AI phrasing everyone’s learned to recognize. Rather than banning AI writing tools outright, Clay set four principles: authorship accountability, meaning a writer has to stand behind every sentence as genuinely their own; writing-as-thinking, on the premise that outsourcing the writing outsources your own understanding of the material; reader respect, since unclear writing imposes a real, multiplicative time cost across everyone who has to parse it; and brevity over length, pushing directly against AI’s tendency toward verbose, vacuous phrasing that should get cut rather than kept.

The policy reportedly went viral both inside and outside Clay after co-founder Varun shared it on LinkedIn — evidence that plenty of other companies are hitting the same AI-writing-quality problem as adoption scales past the experimentation phase.

It lands at a moment when Clay itself is shipping fast enough that internal writing quality has real stakes. Clay’s own account of building Workflows describes its new visual orchestration layer — what the company calls its largest interface shift since the original spreadsheet-style table — complete with version control and audit trails showing exactly which path a given record took through the system. An engineering org shipping that much surface area, that fast, is exactly the environment where sloppy internal writing compounds fastest — and where a rule as simple as “you still own the sentence” earns its keep.