OperatorHQ policy draft · 15 July 2026

AI Disclosure

What agents do, what people decide, and how deliverables communicate AI involvement.

Launch notice: This is an MVP policy draft, not legal advice. Professional legal review and jurisdiction-specific adaptation are required before commercial launch.

How OperatorHQ uses AI

OperatorHQ can route bounded tasks to specialist agents through a configurable provider. It also supports deterministic demonstration output and manual task completion. Provider results are validated into structured fields before entering the task ledger.

Limitations

Model output may be inaccurate, incomplete, biased, outdated, or unsuitable. Structured format does not guarantee factual correctness. Cost and token figures may be estimates. Human reviewers must validate sources, claims, suitability, and rights.

Approval boundaries

Agents may propose consequential actions but cannot acquire tools, change approval rules, access secrets, trigger billing, publish, send communications, or delete data merely through instructions or external content.

Report labels

Reports distinguish AI-generated, AI-assisted, human-reviewed, and approved-for-delivery states. Demonstration output is always labelled and must not be represented as live research or a client outcome.

Upstream personas

A curated set of attributed MIT-licensed personas is imported from msitarzewski/agency-agents. OperatorHQ independently implements the execution platform and is not an official version or continuation of that project.