Monitors

Bind prompts to models on a schedule. Daily and weekly cadences across ChatGPT, Claude, and Gemini.

A monitor is the unit of recurring work. It binds one prompt to one model and one cadence. Each time it fires, it produces a run.

Creating a monitor

Monitors → New monitor asks for three things:

Prompt — pick from the prompts you’ve already created.
Surface — the model you want to test against (ChatGPT, Claude, Gemini).
Cadence — Daily or Weekly. Daily is the default; weekly suits slower-moving categories or larger prompt portfolios on smaller plans.

A single prompt can have multiple monitors — one per model — and that’s the recommended setup. Same question, three answers, a much fuller picture of how your brand shows up.

Pausing and resuming

A monitor can be paused (no new runs created) without losing its history. Use this for seasonal prompts or while you’re iterating on the text via drafts.

Bulk monitor creation

When you ingest a batch of prompts, the ingestion review step offers to spin up monitors on a default surface set in one click. Useful when seeding a new project — start broad on ChatGPT, narrow to specific models once you see which ones matter for your category.

Cost and quota

Each run consumes a credit from your plan’s monthly run budget. The Billing page shows your current usage and a forecast based on configured monitors. If you hit your cap, monitors continue to be scheduled but runs queue rather than fire — upgrading retroactively drains the queue.

API

GET /v1/projects/{projectId}/monitors
POST /v1/projects/{projectId}/monitors
GET /v1/monitoring-surfaces — the catalog of available models / surfaces.

See the API reference for full schemas.