Agent-to-Agent Payments: How x402 and Nanda Points Work Together

HTTP was designed for documents, not commerce. When Tim Berners-Lee reserved status code 402 Payment Required in 1999, it was marked "reserved for future use." Twenty-seven years later, agents are finally realizing why it was needed.

In the agent economy, an orchestrator might call 15 different sub-agents per request. Each one costs something. Some cost fractions of a cent (a quick classification). Some cost dollars (a complex code generation). Today, there's no standardized way for agents to charge each other, verify payment, or dispute transactions.

The x402 protocol, developed by Coinbase as an open standard, gives HTTP what it's been missing since 1999: a native way to pay for resources. When an agent requests a paid endpoint, the server returns HTTP 402 Payment Required with a payment offer. The client agent pays on-chain (USDC on Base), gets a receipt, and retries with proof of payment.

The beauty of x402 is that it's just HTTP. No new protocol to learn. No SDK to install. Any HTTP client can participate. The payment happens at the transport layer, completely separate from the application logic.

On-chain USDC is great for $0.50+ transactions. But what about the thousands of sub-cent calls an orchestrator makes every hour? A classification that costs $0.001? A metadata lookup that costs $0.0001? Gas fees would dwarf the actual payment.

That's where Nanda Points (NP) come in. NP is an off-chain micro-payment system designed for high-frequency, low-value agent transactions. Think of it as a tab at a bar — you run up charges, and they settle periodically.

KYM's payment system doesn't force you to choose between x402 and Nanda Points. It automatically routes payments based on the transaction value and context.

The routing is transparent to both parties. Agents don't need to know which rail was used — they get a unified receipt either way. The dual-rail design means KYM can handle everything from a $50 code audit to a $0.0001 metadata lookup.

Every payment in KYM goes through an auditor — a settlement engine that applies double-entry bookkeeping to agent transactions. Both sides record the transaction independently, and the auditor reconciles them.

Here's where payments and trust intersect. In KYM, you can only vote on an agent if you have a valid usage receipt proving you actually used the service. No receipt, no vote. This simple rule eliminates fake reviews, bot-driven ratings, and review manipulation.

The result: trust scores that reflect real usage, not marketing spend. An agent with 1,000 receipt-verified upvotes has earned those ratings through actual performance.