Know which customers
drove your AI spend.
OpenAI can't tell you which customer drove your bill. We can. Our open-source SDK instruments your LLM calls, tags them to customers, and pipes usage data to Stripe.
AI Provider Invoices
January 2026Your AI bill has no customer breakdown.
OpenAI sends one number. Anthropic sends another. You can't tell which customers drove those costs — or whether your pricing even covers them. Replace your brittle internal logging with infrastructure that works.
Two lines of code. Full attribution.
Wrap your LLM calls with our SDK. We capture tokens, model, cost, and latency — tagged to your customer and feature.
import { Attribute } from '@attribute/sdk'
const ai = new Attribute({
apiKey: process.env.ATTRIBUTE_KEY,
provider: 'openai'
})
// Use exactly like the OpenAI SDK
const response = await ai.chat.completions.create({
model: "gpt-5.2",
messages: [
{ role: "user", content: prompt }
],
// Add these two lines:
customerId: "cus_abc123",
feature: "summarizer"
})
// → LLM call goes direct to OpenAI
// → Telemetry sent async to AttributeInstall the SDK
npm install @attribute/sdk — works with OpenAI, Anthropic, Azure, and Bedrock.
Tag customers and features
Add customerId and feature. We capture tokens, cost, model, and latency automatically.
Connect to Stripe
One-click OAuth. We push metered usage directly to Stripe's billing meters.
Embed customer portals
Drop in our React component. Customers see their usage by feature.
Show customers what they're paying for.
Customers paying for AI expect visibility. Attribute provides embeddable, white-labeled dashboards that live inside your app — usage by feature, limits, and history.
Daily usage
<AttributePortal customerId="cus_abc123" />One component. Shows usage by feature. Themed to your brand.
Your keys. Your calls. Our telemetry.
Attribute is not a proxy. Your LLM requests go directly to OpenAI, Anthropic, or your provider of choice. We only receive usage metadata — never prompts, never responses, never API keys.
Attribution unlocks better unit economics.
Price with confidence
See actual cost-per-customer before setting prices. Move from flat fees to usage-based pricing with data, not guesses.
Catch margin problems early
Identify unprofitable accounts before they drain runway. Real-time visibility, not month-end surprises.
Build customer trust
Customers paying for AI expect to see their usage. Portals reduce billing disputes and churn.
Works with your stack
Free during beta.
We're looking for design partners to shape the product. Early adopters get full access — free — while we figure out pricing together.
Get early access.
We're onboarding design partners now. Join the waitlist and we'll reach out.