← Use cases

AI token and credit abuse

Your AI free tier, for real customers.

Every free request on an AI product runs on paid compute. Per-account limits stop one account from overusing it; they can't see one person running ten accounts. isCredible looks at usage across linked accounts and tells you when to step in.

Free plan with 100 checks a day. No credit card.

The problem

Free-tier farming

Many accounts from one person, each using its full free allowance.

New accounts burning quota

Accounts that use most of their limit within days of signing up.

Sudden spikes

Usage far faster than the account's own pace, often from scripts.

How isCredible helps

  1. 1

    Keep counting usage as you do

    Your app enforces its own limits; isCredible receives the current totals.

  2. 2

    Check before expensive actions

    The SDK only calls the API when a check is due, so checking on every request is cheap.

  3. 3

    Slow down or stop abusers

    Ask for phone verification, offer the paid plan, or stop free usage.

What it catches

  • Linked accounts sharing devices
  • Look-alike emails and shared phones
  • New accounts burning quota
  • Usage spikes
  • Usage over the limit
  • Scripted clients
const { action } = await iscredible.checkUsage(
  user.id,
  { metric: "ai_tokens", used: tokensUsed, limit: 100000 },
  { deviceId, ip: req.ip, accountCreatedAt: user.createdAt }
);
if (action === "DENY") return res.status(429).json({ error: "Free limit reached" });

Built-in scenario SYS_USAGE_SCREENING_001. Also available for Python and any HTTP client.

Questions

How do I stop people from abusing my AI product's free tier?
+
Send each account's current usage with the Verify API or the SDK's checkUsage. isCredible compares it with the accounts linked to it, its age and its own pace, and returns approve, review or deny.
Won't checking on every request add latency and cost?
+
The SDK caches each answer and tells your code when the next check is due, so most requests are answered from memory without an API call.
Which metrics can I send?
+
Any metered resource: AI tokens, API calls, credits, messages, exports or minutes.