My team accidentally left a loop running and spent $500 in two hours. Does Helicone have a way to send Slack or Email alerts if our daily spend exceeds a certain threshold? I need an early warning system to prevent these kinds of budget overruns in the future.
3 answers
Yes, the "Alerts" feature is designed exactly for this scenario. You can create a budget alert based on the total cost over a rolling 24-hour period or a calendar month. When the threshold is hit, Helicone can trigger a webhook. I have ours connected to a small Zapier task that pings our #dev-ops Slack channel and even sends a high-priority alert to PagerDuty if the spike is severe. This has already saved us once when an automated test script went rogue. Beyond just cost, you can also set alerts for "Error Rates," so if OpenAI starts returning 500 errors, your team knows before the customer reports it.
Can we set different alerts for different "properties"? For example, I want a strict budget for my dev team but a larger one for production.
The UI for setting these up is under the "Settings" tab. It’s very intuitive and only takes about two minutes to configure.
Thanks Bruce. I found it! I also noticed you can set alerts for latency, which is going to be great for our SLA monitoring.
Absolutely. You can filter alerts by any custom property you've sent via headers. We use an Environment property to distinguish between Staging and Prod. My staging alert triggers at $50, whereas my production alert is set much higher. It makes it very easy to catch runaway development scripts without causing false alarms during a busy day for our live customers.