Acceptable Use Policy

Purpose

This Acceptable Use Policy (“AUP”) supplements the Terms of Service. Where the Terms set the contractual frame, this AUP describes the operational envelope inside which the Service is designed and tested to perform. It exists to keep customers, the grids they operate in, and the public safe from misapplication of forecast outputs.

In-scope use cases

• Day-ahead and short-term load forecasting for the U.S. ISOs and RTOs documented as supported in the API reference.
• Forecast horizons and resolutions documented in the API reference (1-hour or sub-hourly resolution where available; horizons scoped by grid, plan, and validation status).
• Operational planning, scheduling, market-bidding decision support, and research analytics that include human review of forecast outputs.
• Customer-internal dashboards, reporting, and analyst workflows where the forecast informs but does not solely automate decisions with material financial or safety consequences.

Out-of-scope use cases

The Service is not designed, tested, or warranted for these uses, and customers must not deploy it for them without independent validation, redundant feeds, and explicit human-in-the-loop approval:

• Safety-critical automation. Direct control of grid dispatch, breaker operations, automatic generation control, or any closed-loop safety system.
• Sub-minute trading control loops. Tick-level algorithmic trading, high-frequency arbitrage, or automated position changes triggered without independent risk checks.
• Markets outside our published coverage. Use in non-U.S. ISOs, in commodity markets other than electricity, or for fuels not listed in the API reference is not a supported use case and forecast quality is not warranted.
• Sole basis for material trades or operational commitments. Customers must maintain at least one independent forecast feed (whether in-house, ISO baseline, or third-party) for any decision with material financial or operational consequence.
• Resale or redistribution. Forecast outputs may be used to inform Customer’s own decisions and internal products, but may not be packaged, syndicated, or sold to third parties without a separate Data License agreement.

Human-in-the-loop expectations

For any use case where a forecast error could result in material financial loss, regulatory exposure, grid-reliability risk, or harm to persons or property, Customer must have a designated reviewer responsible for sanity- checking forecast values against an independent reference before action is taken. Customers operating against the API in a fully unsupervised mode do so at their own risk and outside the scope of warranty.

Pipeline status and warming-up signals

The Service publishes operational signals (forecast freshness timestamps, the “warming up” state on the live dashboard, the public health endpoint at /api/v1/health, and per-region issue times). Customers integrating against the Service in automated workflows should consume these signals and refrain from acting on forecast values flagged as stale, warming up, or otherwise degraded.

Updates to this policy

We may update this AUP when operational guidance, regulatory requirements, or supported markets change. Material updates are announced 30 days in advance via email and the dashboard. Continued use of the Service after the effective date constitutes acceptance.

Contact

Questions about whether a specific use case is in scope: hello@gramm.ai. For Enterprise deployments, request a call so we can confirm scope-of-fitness in writing.