Ant's car agent now pays tolls and fills the tank in 16 million cars
Shanghai's Inclusion·Bund Conference closed on Friday with China's agents doing the boring things — paying for fuel, tolls and movie tickets — and the numbers attached were bigger than the demos.
Ant Group's in-car agent just took over the parts of driving that involve money. At the conference (Sept 9–12), Alipay launched three new cockpit capabilities for its Abao service agent — voice-activated highway toll payment without stopping or handing over a card, in-car movie ticket booking from search to checkout, and one-tap refueling that finds the nearest station and settles the bill. Ant says the services already reach 17 mainstream automakers and more than 16 million connected cars, with 60-plus newly signed OEM partners. The plumbing is the interesting part: Ant's AHA cross-device protocol lets a task begun in the car finish on the phone after you get out, and a merchant's single service adaptation is distributed to phone, head unit and AI glasses at once. China's agents are being graded on transactions completed, not conversation quality — which is the same argument we saw from the vendor side last week in Alipay's payment robot dog went shopping — agentic commerce gets its first legs.
At the same conference, embodied-AI people quietly changed the metric. Ant LingBo chief scientist Shen Yujun argued the VLA-versus-world-model debate is a detour: what matters is whether a robot can extract usable representations from continuous vision and touch, align those modalities, and emit actions that actually execute. His two tests for a "general brain" were deployment cost falling with each model upgrade, and a robot surviving a full day in a live conference hall without incident — not nailing one scripted demo. Chief data scientist Huang Youtao said the data flywheel isn't spinning because too few robots are in real production jobs, so failures never flow back, and SenseTime co-founder Wang Xiaogang set a commercial bar: thousands of robots in one scenario and a customer at break-even. One shipped example: a LingBo-brained robot already sorting prescriptions in a Guoda Drugstore pharmacy through an 80 cm aisle, no shelf retrofits. Ant also opened an embodied-model competition on LingBot-VLA 2.0 — the trust-and-tooling layer agents need before they can act on our behalf is exactly what we tracked in Visa, Mastercard and Ant want one trust layer for AI agents.
What to watch: whether any Western payment network or automaker ships a comparable voice-pay contract this quarter, and whether 16 million becomes a number anyone else can quote.
Ant is selling in-car agents on toll booths and gas pumps; OpenAI is selling them on keyboard clicks. Which version do you actually want listening in your car? Tell us in the comments.
Sources: Leiphone · MyDrivers · Leiphone — embodied AI forum · Xinhua
