14-day free trial • No card needed to start • 2-minute setup
Your API keys stay with you. We only see aggregated spend data — never prompts or responses.
GitHub Action runs hourly. No SDK, no code changes, no maintenance.
No dashboards to check. Get notified in email or Slack — only when something needs attention.
OpenAI and Anthropic. Your app's AI spend in one place.
We learn your baseline and notify you when something looks unusual.
Define your budget. We'll notify you as you burn through it — so you're never caught off guard.
One click. We'll create your DriftInsight account and API token.
Choose your method: GitHub Action, Cloudflare Worker, or cron script.
Insights start with the first sync. Anomaly detection improves as we learn your baseline.
Real notifications — delivered to your inbox or Slack
Click the tabs above to preview different notification types
One plan. Everything included. Cancel anytime.
or $49.99/year (save $10)
No card needed to start your trial
DriftInsight tracks your application's AI API spending for OpenAI and Anthropic, without requiring any changes to your code. We read usage data directly from your provider's billing API and notify you when something unusual happens: sudden cost spikes, unexpected model usage, or volume drops. We also alert you as spending approaches your budget thresholds. Set up a GitHub Action once, and we handle the rest.
Provider dashboards show you what happened—you have to remember to check them. DriftInsight learns your normal spending patterns and proactively notifies you when something changes or when you're approaching your budget limits. Costs spiked overnight? Volume dropped (maybe your AI feature broke)? Hitting 80% of your monthly budget? You'll know without logging in anywhere.
Most AI observability tools (Langfuse, Helicone, LiteLLM, etc.) require you to route traffic through their proxy or add SDK calls to your code. That's great for teams who need tracing and debugging, but it's overkill if you just want to avoid surprise bills.
DriftInsight works differently: we read your usage data directly from your provider's billing API. No proxy, no SDK, no code changes. Your API calls go straight to OpenAI or Anthropic as they always have. We just monitor the spend and alert you when something's off.
If you want deep tracing and prompt debugging, use the heavier tools. If you just want simple budget alerts without touching your code, that's us.
Google's Gemini API doesn't currently offer a billing/usage API endpoint that third parties can query. Without programmatic access to your usage data, we can't collect it. We're watching for changes and will add Gemini support as soon as Google provides an API for it.
We don't expect you to check charts regularly—you won't. Instead, we stay quiet when everything's normal and only reach out when there's a real signal worth your time.
We have an account portal where you can set up your data collectors, configure notification preferences (email and Slack), set budgets, and see a quick summary of your account activity. But the portal is for configuration, not monitoring. The monitoring comes to you—via email or Slack—so you don't have to form a habit of checking yet another dashboard.
We deduplicate notifications—you'll only receive one notification per type, per provider, per day. If we already notified you about a cost spike for OpenAI today, we won't send another one until tomorrow. We also use tiered thresholds: early in your account's life (when we have less data), we're more conservative and only notify on larger anomalies. As we learn your patterns, detection becomes more precise. The goal is signal, not noise.
You don't need to check anything. We notify you when action is needed.
Yes. Every new account gets a 14-day free trial with full access. No credit card required to start. If you don't upgrade, we'll remind you a few times, then your data syncing stops—but we keep your historical data for 30 days in case you change your mind.
We store only aggregated daily usage data: total cost, request counts, and model breakdown per day. We never see your prompts, responses, or API keys. The collector runs in your own GitHub Actions, reads usage data from your provider's billing API, and sends us only the summary. Your API key stays in your own secrets.
No. We use Lemon Squeezy (a Stripe company) as our payment processor. Your card details go directly to them—we never see or store your payment information. Backed by Stripe's payment infrastructure, Lemon Squeezy is fully PCI-DSS compliant and handles all payment security and global tax compliance. We only receive confirmation that your subscription is active.
Yes. You can cancel from your account page at any time. Your access continues until the end of your billing period, and you won't be charged again. No questions, no hoops.
You need:
That's it. No SDK to install, no code changes to your app. Sign up with GitHub, set up a GitHub Action with your DriftInsight token and provider API keys, and you're done. You'll get your first spend snapshot within the hour.
The collector supports multiple providers in a single sync. Configure all your API keys in the GitHub Action, and we'll track each provider separately. You'll see breakdowns by provider in your notifications and can set per-provider budgets.
Join teams who know exactly what they're spending on AI — and why.
Get Started Free →