What’s the recommended way to get notified when my enrollment’s AIP capacity is reaching a certain threshold e.g. 80%? While I’ve configured default max rate limits to where I want it for my projects, it would be great to still get notified if we get to a point where we are reaching a certain threshold for usage so we can investigate what’s causing that amount of usage.
There is no rate limits alerting feature at the moment in AIP. The closest option is to use the AIP usage and limits tool in Resource Management App ((https://www.palantir.io/docs/foundry/resource-management/llm-capacity-management/#view-usage)) and apply the rate limits toggle (top right) to see when you’ve hit limits, and then zoom in on that time frame and look into the specific projects and resources.
Once you’ve diagnosed the problem, you have several levers to pull to address the problem: schedule pipelines to different times of day; diversify which models you use; request an enrollment tier upgrade or Reserved Capacity; adjust Project Rate Limits.
If the existing set of tools is insufficient, please let us know with any additional details that you’re able to share, we’re always happy to hear from our users!