How Lava Works
Three Simple Steps
Lava handles the full billing loop for AI-powered applications in three steps:
- Credits Issued — Your customer loads their wallet with prepaid credits via Lava's embedded checkout. Funds are held securely and ready to spend.
- Usage Tracked — As your application makes AI API calls through Lava's proxy, every token and request is metered in real time with sub-cent accuracy.
- Payouts Sent — Revenue from usage is automatically calculated, and payouts are sent to your bank account on a regular schedule via Stripe Connect.
The Transparent Proxy
At Lava's core is a transparent API proxy. Instead of calling AI providers directly, your application routes requests through Lava. The integration is just two lines of code:
- Change the base URL to Lava's proxy endpoint
- Add a Lava authorization header
Lava forwards the request to the AI provider (OpenAI, Anthropic, Google, etc.), streams the response back to your user, and meters the usage — all with less than 20ms of added latency.
Pre-funded Wallet Model
Lava uses a prepaid wallet model, not post-paid billing. Customers fund their wallet before using services. This means:
- No surprise charges — Users control their spending
- Guaranteed payment — Merchants are paid from pre-funded balances
- Real-time balance — Users always know exactly how much credit they have left
Who Uses Lava?
Lava is designed for two types of users:
- AI developers and SaaS companies who want to monetize AI-powered features without building billing infrastructure from scratch
- Individuals and teams who want consolidated AI spending with per-tool budget controls via AI Spend keys