4亿。
这是支付宝“碰一下”功能上线两年后交出的用户成绩单。基于线下超过3000万物理点位的AI化升级，“碰一下”在2026年将重点开放Agent服务网络。这意味着支付宝不再只是一个“碰一下付款”的支付工具，它正在变成中国最大的线下AI服务分发网络。
从支付到Agent：支付宝的AI进化路径
回头看，“碰一下”的演进路径非常清晰。2024年7月上线时，它解决的核心问题是支付体验——掏出手机碰一下POS机就能完成付款，比扫码更快更方便。但支付宝的野心从来不是做一个“更快的付款方式”。
3000万线下物理点位是支付宝手里最值钱的资产。便利店、餐厅、加油站、地铁站、自动售货机——这些点位每天发生着数以亿计的交互。过去这些交互的终点都是“付款成功”，但如果在付款的同时，AI Agent能帮你完成更多事情呢？比如：碰一下咖啡店的点位，自动点你常喝的那款咖啡并积分；碰一下加油站，自动匹配最优加油优惠；碰一下景区门口，直接AI导览加购票。
这就是Agent服务网络的本质：把每个物理点位变成一个AI Agent的触发入口。手机碰一下的瞬间，背后的Agent识别场景、调取用户偏好、执行服务流程、完成支付闭环。2026年ChinaJoy上，支付宝已经首发了“碰一下组队”“积分碰一碰”等Agent化功能。
国民应用的Agent战争还在继续
支付宝不是唯一在下这盘棋的玩家。6月8日微信正式开放AI Agent生态内测，首批接入美团、京东、滴滴等头部服务平台；6月16日支付宝重构App底层，推出原生AI助手“阿宝”。两大国民应用在5天内先后亮剑，16亿人的屏幕正在变成Agent的对话界面。
但支付宝“碰一下”有一个微信短期内难以复制的优势：物理世界的入口。微信的强项是线上社交和服务生态，但线下场景的密度和深度，支付宝依托3000万物理点位和多年的商户关系，有明显的先发优势。当Agent服务网络全面开放后，每个商户都可以在自己的点位上部署定制化的Agent服务——这本质上是线下版的“小程序生态”，只不过入口从“扫码”变成了“碰一下”，交互方式从“点击菜单”变成了“和Agent对话”。
WAIC期间，支付宝还与阶跃星辰宣布达成AI Agent系统级合作，围绕STEP-X原生AI终端展开深度协同，用户通过自然语言即可调用阿宝服务完成跨端操作。这意味着“碰一下”的入口不局限于手机，未来可能扩展到各种AI原生硬件终端。
4亿用户、3000万点位、开放Agent网络——支付宝的AI战略正在从“App内加个AI助手”升级为“构建物理世界的Agent基础设施”。明天见。
400 million.
That's the user count for Alipay's “Tap-to-Pay” (Peng Yi Xia) feature two years after launch. Building on AI upgrades across over 30 million physical offline touchpoints, Tap-to-Pay will focus on opening its Agent service network in 2026. This means Alipay is no longer just a “tap to pay” tool — it's becoming China's largest offline AI service distribution network.
From Payments to Agents: Alipay's AI Evolution
In hindsight, Tap-to-Pay's evolution path is clear. When it launched in July 2024, its core problem was payment experience — tap your phone on a POS terminal and pay, faster and more convenient than scanning QR codes. But Alipay's ambition was never to be a “faster way to pay.”
Those 30 million offline physical touchpoints are Alipay's most valuable asset. Convenience stores, restaurants, gas stations, subway stations, vending machines — hundreds of millions of interactions happen at these points daily. Those interactions used to end at “payment successful,” but what if at the moment of payment, an AI agent could do more? Tap a coffee shop's point and automatically order your usual with loyalty points; tap a gas station and auto-match the best fuel discount; tap a scenic area entrance and get AI-guided tour plus ticket purchase in one go.
That's the essence of the Agent service network: turning every physical touchpoint into a trigger for an AI agent. The instant your phone taps, the agent behind it identifies the scenario, pulls user preferences, executes the service flow, and closes the payment loop. At ChinaJoy 2026, Alipay already debuted agent-powered features like “Tap to Team Up” and “Tap for Points.”
The Super-App Agent War Continues
Alipay isn't the only player making moves here. On June 8, WeChat officially opened its AI Agent ecosystem beta, with first-wave integrations including Meituan, JD.com, and Didi; on June 16, Alipay rebuilt its app foundation and launched native AI assistant “A Bao.” The two national super-apps drew their swords within five days of each other, and 1.6 billion people's screens are becoming agent conversation interfaces.
But Alipay Tap-to-Pay holds one advantage WeChat will struggle to replicate short-term: physical-world entry points. WeChat's strength is online social and service ecosystems, but in offline scenario density and depth, Alipay has a clear head start through its 30 million touchpoints and years of merchant relationships. When the Agent service network fully opens, every merchant will be able to deploy customized agent services at their own locations — essentially an offline version of the “mini-program ecosystem,” except the entry point shifts from “scan QR” to “tap,” and the interaction shifts from “click menus” to “talk to an agent.”
During WAIC, Alipay also announced system-level AI Agent cooperation with StepFun (Jieyue), integrating deeply around the STEP-X native AI terminal, where users can invoke A Bao services through natural language for cross-device operations. This means Tap-to-Pay's entry point won't be limited to phones; it may expand to various AI-native hardware terminals in the future.
400 million users, 30 million touchpoints, an open Agent network — Alipay's AI strategy is upgrading from “adding an AI assistant inside the app” to “building agent infrastructure for the physical world.”
See you tomorrow.
每个物理点位都是Agent的触发入口——这是线下版的小程序生态，只不过交互从“点菜单”变成了“和AI对话”。
—— Dawn Vision编辑部
Every physical touchpoint is an agent trigger — it's an offline mini-program ecosystem, except interaction shifts from “clicking menus” to “talking to AI.”
— The Dawn Vision Editorial Desk
Alipay · Tap-to-Pay · 400M users · Agent service network · A Bao · 30M touchpoints · offline AI · WeChat-Alipay agent war
Sources · 信源 Sources
- 新浪科技 - 支付宝“碰一下”用户数破4亿 将重点开放Agent服务网络
- 36氪 - 阶跃与支付宝达成AI Agent系统级合作
- 今日头条 - 支付宝碰一下用户破4亿，下一步增长要靠AI
本文基于 Dawn Vision 认知引擎处理的 9 个源信号生成，经编辑部人工审核。素材来源：新浪科技、36氪、今日头条。
This article was generated by the Dawn Vision cognitive engine processing 9 source signals, with human editorial review. Sources: Sina Tech, 36Kr, Toutiao.
