Agent platforms

Paying paywalled APIs at inference time

Your agent calls a data enrichment, search, or LLM inference endpoint and gets back 402 Payment Required. Today you either pre-purchase credits (overpay), proxy through a custodial gateway (compliance risk), or hard-code a single rail and break when the endpoint changes its accepted protocol.

Routeweiler handles the 402 transparently, routes to x402 on Base USDC by default, and retries. Your agent sees only the final 200. No credits, no custodian, no glue code.

x402 recipe →
LLM & inference services

On-chain billing without a payment gateway

You want to monetise inference calls with stablecoin micropayments but don't want to build and maintain a payment-rail integration inside your inference server.

Your client agents install Routeweiler; your server issues a standard 402 with an MPP-Tempo header. Routeweiler pays the Tempo stablecoin transfer, you receive the settled amount, and a signed trace receipt lands in both parties' local SQLite for reconciliation.

MPP-Tempo recipe →
Finance & ops teams

Reconciling and capping autonomous agent spend

You're operating a fleet of AI agents and need to cap spend per agent, per session, or per rail, and then reconcile what was actually paid against what was authorised. Existing solutions either require a hosted billing service or trust the agent to self-report.

BudgetEnvelope enforces a hard cap in local SQLite with Ed25519-signed draw receipts. The agent physically cannot overspend the envelope. Every payment emits a TraceEvent with rail, amount, and settlement proof, giving you a tamper-evident audit log with no hosted backend required.

Trace & budget docs →