Alipay Just Gave a Billion Users an AI Agent. Here's What It Can Do.

Alipay — the payments super-app that processes more transactions than Visa and Mastercard combined — just opened its AI agent to everyone. No invite code. No waitlist. iOS and Android users in China can search "Abao" or "蚂蚁阿宝" right now and start using it.

This is a big deal. Not because the technology is revolutionary, but because of the scale at which a real, action-taking AI agent just shipped.

What Abao Actually Does

The core idea is simple: instead of navigating Alipay's sprawling menus to find what you need, you just say it.

Tell Abao "查公积金" (check my housing fund) and it finds the right mini-program, routes you there, and surfaces the service. It understands intent and takes action — not just answering your question, but doing the thing.

That's the definition of an agent. It's not a chatbot that spits out information. It's a system that understands what you want to accomplish and moves you toward it.

The Safety Line They've Drawn

Ant Group was careful about one thing: money doesn't move without you.

All financial actions — payments, transfers, QR code scanning, any transaction — require explicit user confirmation. Abao can navigate you to the right place and queue up an action, but your finger hits the final button. That's a smart constraint, and probably a legal one.

This matters because it draws a clear distinction between orchestrating a task (AI's job) and authorizing it (your job). It's the same architecture that most well-designed agent systems use: the agent proposes, the human approves.

Why This Moment Is Different

AI agents have been shipping in productivity tools for a while — in coding assistants, in customer support, in research tools. But those are still mostly power-user products. Alipay is something else: it's a utility. Hundreds of millions of people use it daily for rent, groceries, transit, utilities.

When an AI agent ships there, it stops being a feature for the tech-curious and becomes infrastructure for everyday life.

The transition point isn't when agents become smart enough. It's when they become embedded enough. Alipay just cleared that bar.

The Chinese Super-App Advantage

One reason this works in Alipay specifically is the nature of the app itself. Alipay isn't just payments — it's thousands of mini-programs, services, and government portals all living inside a single ecosystem. It's a platform with deep integrations across the kind of things people actually need to do: check benefits, pay bills, file paperwork, book appointments.

That's an ideal environment for an agent. The more integrations exist, the more useful the agent becomes. When every service is already inside one app, "do X for me" becomes genuinely tractable — you're not hopping between apps, you're navigating a connected graph of services with an AI doing the routing.

What This Means If You Use OpenClaw

The architecture Alipay is using with Abao — intent understanding, tool routing, human confirmation on high-stakes actions — is the same pattern that powers OpenClaw.

OpenClaw agents work by taking a natural language goal and figuring out which tools to invoke, in what order, with what inputs. The human stays in control of what actually happens. The agent handles the planning, routing, and execution up to that boundary.

The difference is that OpenClaw is for your workflows, not Alipay's. You define the tools your agent has access to, the tasks it should handle, and the context it carries between sessions. Whether that's checking a database, triggering a report, or monitoring a process — you're building the same kind of action layer that Abao provides inside Alipay, but for whatever you actually need to automate.

What Alipay proves is that people will use agents when they're embedded in things they already rely on. That's the bet OpenClaw is built on too.

Start your free trial →