K (base.eth)

Tab update, and where this is going. I'm updating the website this week. The dashboard is rebuilt and the landing page is getting the full payment walkthrough. That's the small news. The bigger news is what tab is for. Right now an AI agent that wants to buy something has two options: hold a wallet it can drain, or ask a human every time. Both are wrong. Tab is the third option. The owner sets the limits in a contract, the agent signs, the chain enforces. No facilitator, no server, no bug in the agent can spend past what the owner wrote down. That changes what an agent can be trusted to do on its own. Here is the roadmap, in the order I'm building it: 1. Any agent, any API. SDK and MCP server on npm, and the standard x402 facilitator API so every existing x402 seller can accept a tab without changing their code. The goal is that paying for an API call from an agent is one line and the human only ever sets a cap. 2. Agents hiring agents. A seller is any address that answers 402, so two agents with tabs can already pay each other. Next is making that the default: an agent with a tab can subcontract work to another agent, and both owners keep their limits. That is an economy between agents where every participant has a spending policy a human wrote. 3. Sub-cent AI. A fee that covers gas and batched settlements so one transaction carries many payments. Then a model call can cost a fraction of a cent and still settle on chain before the response is served. 4. Tabs in stocks. Robinhood Chain has tokenized equities and the vault accepts any ERC-20. An agent can be given a tab funded in a stock token, spend it on compute or data, and the owner keeps the exposure they chose. Money for agents that isn't only stablecoins. 5. Tab as the standard. The policy vault as the way any wallet gives any agent money: mint a key, set a cap, revoke when done. If that becomes the default, no agent needs to hold funds again