How do I get Slack alerts when AI spend spikes?
Connect a Slack channel with an incoming webhook and get budget-threshold, anomaly and security alerts as they happen. Base plan and above.
Goal: a Slack message when a budget crosses a threshold, spend spikes, or a key starts behaving like it was compromised. Plan: Base and above. Time: about 2 minutes.
Step 1 — Create an incoming webhook in Slack
- Open api.slack.com/apps and create an app From scratch in your workspace (any name, e.g. "Cognocient").
- In the app's settings, open Incoming Webhooks, switch it on, click Add New Webhook to Workspace and pick the channel.
- Copy the webhook URL (
https://hooks.slack.com/services/...).
Step 2 — Paste it into Cognocient
Open Settings → Alerts, paste the URL and click Connect & send test. Cognocient posts a test message first and only saves the URL if Slack accepts it. The URL is stored encrypted and never shown again.
Step 3 — Choose what is posted
| Switch | What it posts | Default |
|---|---|---|
| Budget thresholds | Each alert threshold a budget crosses (50%, 80%, 100% by default), once per budget per month | On |
| Anomalies & security | Spend and call-frequency anomalies, rejection-rate spikes, velocity-limit trips, compromise-risk signals, shadow-spend gaps | On |
| Executive anomaly summaries | A plain-English version of each spend anomaly for leadership | Off |
| Report deliveries | A note when a scheduled AI ROI Summary has been emailed | On |
Quiet hours suppress posts during a daily window in the timezone you choose. Send test message checks the connection at any time.
Step 4 — Check it end to end
Create a budget with a very small limit (e.g. $0.05) on a test feature, make a few calls tagged with that feature, and watch the channel. Delete the budget afterwards.
Related articles
Tag Your First AI Call
Add 2 headers to your existing code and see per-feature spend in under 5 minutes.
Set a Monthly Spending Limit
Create a hard budget enforced at the proxy before charges reach your provider bill.
Cut Your AI Bill with One Click
Use AI Advisor recommendations to apply model downgrades and caching without code changes.