Keyban will be in the house! 😎 Pitching our Agent Wallet solution at the Silicon Valley Summit. See you there! 🚀

Thanks for sharing these info. Pay-per-call opens up useful services that subscriptions don't always suit. Would love to see repeat usage and unique paying wallets alongside transaction counts to better understand adoption. And what may be main destinations...

Congrats on the launch 👏 RPC access via x402 is a practical building block for agents. Pairing this with per-agent budgets gives teams a useful combination: autonomous access to infrastructure and predictable spending.

Pay-per-tool makes every MCP call an economic decision. The toolstore can quote a price, but the agent’s wallet still has to check whether that provider and amount fit the workflow budget — and request approval when they don’t. This is what we develop...

Keeping the payment stack intact makes sense. Agents need authorization before checkout and a complete audit trail showing what was requested, which rules were applied, what was approved and where any error occurred.

Identity tells the counterparty who stands behind the agent. The company still needs to define what that agent may do with money (budget, approved payees, transaction limits etc) and when a human approval is required... before the wallet signs.

Usage visibility is useful, but finance also needs a control interface to set the rules about total budget, approved services, payment limits and approval thresholds... Seeing spend after the fact isn’t the same as controlling it before payment.

An agent that can pay is useful. An agent that knows when not to pay is deployable. Keyban Agent Wallet lets builders enable payments while finance sets budgets, payees, limits and approval thresholds. Start the trial: keyban.io/en/agent-wallet/

With real capital at stake, completing the task isn't enough. The agent must stay within budget, pay approved recipients and request approval when needed.Keyban gives builders the wallet and finance teams the controls. Give it a try: keyban.io/en/agent-wallet/

Agent-to-agent payments make the authorization question even more interesting. Can the first agent pass part of its budget to the second, and what happens when the final batch exceeds the original mandate?

Putting real capital at stake changes what should be measured. An agent should be evaluated not only on its result, but also on whether it stayed within budget, used allowed counterparties and requested approval when required #My2Cents.

Task markets are where agent wallets get real: an agent hiring other agents needs a budget, not a card. We already see a few x402 services that both earn and spend. That loop is the interesting part.

Interesting split we saw on-chain: x402 on Solana looks like value payments (gift cards, prepaid cards) with mostly one-off buyers, while Base is agents hitting APIs for fractions of a cent. Also, apparently few Solana x402 payers we checked used a smart wallet...

Who's actually buying? Over the last 30 days we counted ~170 distinct wallets paying @nansen_ai via x402 on Solana. All plain keys, zero smart wallets. A small group of agents making lots of 1-cent calls. The next unlock isn't price, it's per-agent managed budgets.

With x402, every inference call can become a payment. A funded wallet isn’t enough: agents need budgets, per-request limits and approval rules outside the model. Give your agent a budget, not a blank cheque. Try Keyban: keyban.io/en/agent-wallet/

Exactly. It’s sad that x402 isn’t supported here for managed wallets. Agents should be able to make microtransactions without being given unrestricted access to funds — with budgets, limits and approvals controlled by the wallet owner.

Pay-per-request inference is a real use case for agent wallet. Builder enables x402 payment. Finance still needs to control providers the agent can pay, amount to spend and when approval is needed. Keyban Agent Wallet is built for. Want to give a try? keyban.io/en/agent-wallet/

Before deciding who pays for a bad purchase, we should prevent it. Our Keyban Agent Wallet lets finance teams control budgets, payees, limits and approvals before the agent pays. Builders can implement the agent but finance defines what it is allowed to do.

Dynamic, yes. Unlimited, no. Let the agent renegotiate but within a preset envelope... and escalate anything beyond it to a human decision. That's the difference between autonomy and a blank check.

Building the agent is the easy part now. The hard part: what is it allowed to spend, who's liable, how do you prove it? Not only ours but also @Mastercard Opinion. That's infrastructure: Keyban Agent Wallet. Want your agents to buy safely? cc @pymnts pymnts.com/commerce/ecommerc…