Appearance
Quotas, limits and credits
AI costs money, so RelyAI gives you three levers to keep it in check: rate caps that throttle bursts, an optional daily cap per customer, and the option to move usage onto customers' own provider keys. This guide walks through each.
Per-task rate caps
Open RelyAI → Settings → Quotas & Limits. Each AI task has a rate cap so no one can hammer your provider. The defaults are 20 per minute and 500 per hour. Lower them if you want a tighter leash; raise them if legitimate use is being throttled.
RelyAI → Settings → Quotas & Limits
On top of your own caps, RelyAI respects your provider's rate limits: if the provider returns a "too many requests" response (a 429), RelyAI backs off rather than pushing through.
Per-customer daily cap
You can also set an optional per-customer daily cap — the maximum AI requests a single customer can make in a day. The default is 0, which means uncapped. Set a number to stop any one customer from consuming a disproportionate share.
AI Credits
If your plan meters AI, customers see an AI Credits card on their subscription page, sitting alongside the Sending and Verification credit cards. It shows how much is remaining this cycle. On plans without metering, the card reads No AI metering on this plan. These credits come from AcelleMail's own AI-generation credit type, so they follow the plan you've already configured there.
Moving usage onto customers' own keys
If you'd rather not spend your own provider budget on customer activity, let customers bring their own key. Enable that option when you connect an engine; customers then enter a key under Account → AI, and their AI usage runs on their provider quota instead of your system key. This is the cleanest way to make heavy users cover their own cost.