Inside Shanghai's Inclusion Conference on the Bund: AI Becomes an Everyday Engine

By Neha Mehta  |  Founder, FemTech Partners · Author, One Stop

I've sat through a lot of AI conferences that talk about the future. This one, the 2026 Inclusion Conference on the Bund, mostly talked about right now.

Walk the 15,000-square-metre show floor here in Shanghai and AI doesn't announce itself with a keynote slide. It shows up as a coffee order placed before you've woken up, a robot squeezing down an 80-centimetre pharmacy aisle at 2 a.m., a tree planted in a desert 3,000 kilometres away by someone who's never seen it in person. Across more than 40 forums and 300-plus exhibiting companies from over 50 countries, the theme organisers chose — “Building the AI Economy Together” — turned out to be less a slogan than a fair description of what I actually saw.

From Generating Text to Executing Tasks

The clearest framing I heard for this moment came from a session with Philippe Aghion, the Nobel laureate economist, who used the conference stage to lay out what he sees as the real test facing China's AI industry: not whether the models are good enough, but whether a technological advance can become a genuine economic one — diffusing out of a handful of high-growth tech firms and into manufacturing, healthcare, and the vast small-business economy that actually employs most people.

The macro numbers support the urgency. China's digital industry generated 20.71 trillion yuan (roughly $2.8 trillion) in revenue in the first half of 2026 alone, up 13.6% year-on-year, with profits growing faster still, at 19.3% — gains regulators partly credit to domestic AI demand. Meanwhile, large language models are now being invoked at a scale measured in the hundreds of trillions of tokens a day, which is another way of saying AI has quietly graduated from a developer sandbox into everyday business infrastructure.

Aghion's own framework — creative destruction — is a useful lens here. New technology doesn't lift every firm equally; it accelerates a sorting process, rewarding fast movers and squeezing out the rest. He estimates AI-driven automation could add roughly 0.7 percentage points to annual productivity growth over the next decade, with more upside still if it reshapes how new ideas themselves get produced, not just how goods get made. He was also unusually direct in praising China's frontier position on measures like high-tech patent share, arguing that Europe's problem now isn't just catching up to the US — it's catching up to China too.

What I Actually Saw on the Floor

For someone whose work lives in payments and financial inclusion, the most telling exhibits weren't the flashiest ones.

At Shanghai's Guo Da Drugstore, delivery robots now navigate those narrow aisles overnight to sort and hand off medicine, one of more than 40 embodied-AI systems on display — the goal, exhibitors told me, is to take the repetitive load off night-shift pharmacists, not replace them. Nearby, Ant Group's health app AQ, which just crossed 150 million users, is doing something quieter but arguably more important: turning AI from a chatbot you consult into a habit you keep, tracking sleep, nutrition and weight for an entire household and nudging people toward better routines rather than just answering medical questions when asked.

Then there was the detail that got the most genuine laughs on the floor: a “Coffee Alarm Clock,” built jointly by Starbucks and Alipay's AI agent Ah Bao, that orders the same iced Americano every morning at 10 a.m. without being asked twice. It sounds trivial until you look at the scale behind it. Since launching in June 2026, Ah Bao has delivered more than 10,000 AI-powered everyday services across more than a dozen categories — paying utility bills, booking pet services, finding EV charging stations, even hailing a self-driving car — and retail brands including McDonald's, KFC, Mixue, Luckin Coffee and Haidilao now let customers order and pay just by chatting with the agent. Its AHA protocol, built for cross-agent and cross-device interoperability, is stitching that same functionality into AI glasses, smartphones and cars: Ah Bao is already integrated with five of the leading smartphone brands, together covering more than 70% of the market, and 16 automakers, so a Xiaomi phone owner can simply ask their own assistant, Xiao AI, to hand a task straight to Ah Bao. Once a user authorises and confirms a request, the agent completes the payment on its own — no separate app, no repeated approvals. That is the shift from an assistant that talks to one that acts, and it is exactly the shift that raises the questions I care about most: how those spending limits get set at the point of authorisation, and who is liable if something still goes wrong. China's payment companies are building that rail-and-consent infrastructure in public, in real time, which makes this one of the more instructive live experiments anywhere for agentic commerce.

The Sustainability Story I Didn't Expect

The exhibit that stayed with me longest, though, had nothing to do with payments.

Ant Forest — the tree-planting mini-program tucked inside Alipay that just turned ten — marked its anniversary at the conference with an open letter to the more than 700 million people who have used it over the past decade. Of those, 160 million have gone on to actually plant trees: over 700 million of them, across deserts in Gansu, Inner Mongolia, Ningxia and Xinjiang, covering an area roughly half the size of the Ulan Buh Desert, China's eighth-largest.

On stage, Ant Forest hosted the finals of its Green Tech Innovation Competition, part of a new “Earth Partner Program” run jointly with Green Technology Bank. Two hundred and ten youth teams from around the world had spent months building solutions across three tracks: desertification control and water access in arid regions, biodiversity protection, and ocean conservation. Ten finalists — vetted by judges including a Chinese Academy of Engineering member and a former UN Environment Programme chief scientist — competed for grants of up to 300,000 yuan, with the most viable projects earning a shot at piloting on real Ant Forest land at Green Technology Bank's experimental base in Xinjiang.

One of the judges, desert-ecology researcher Lei Jiaqiang, made a point I found more interesting than any single pitch: AI, he noted, is now being applied directly to sand and desertification control itself — not just described on an exhibition floor, but actually put to work fixing dunes and managing scarce water in the field.

Where I Think This Actually Lands

None of this reads to me as a finished story, and I don't think it should be presented as one.

The optimistic case is real: AI is diffusing into pharmacies, households, small merchants and reforestation projects in ways that look a lot like the access-to-outcomes shift I've spent my career arguing fintech needs to make. But Aghion's own caution is worth repeating alongside the good news, because it applies as much to payments and financial inclusion as it does to chatbots. He pointed to the late-1990s US tech boom, where early productivity gains from firms like Microsoft and Amazon eventually gave way to rising market concentration, falling rates of new business entry, and stalling innovation — and he warned that the AI value chain, especially cloud computing and GPUs, is already concentrated in a similar way. His prescription wasn't more regulation for its own sake, but institutional adaptation: competition policy that keeps pace with the technology, education systems that teach people how to learn rather than what to know, and labour-market safeguards — he pointed to Denmark's “flexicurity” model — that cushion the transition rather than ignore it. More than 200 economists and AI researchers have separately called for exactly this kind of institutional groundwork.

For fintech ecosystems watching China from the outside, including India's, that's probably the more useful takeaway than any single robot or app: the technology is arriving fast. Whether it lifts the many or concentrates around the few is still, genuinely, an open question.

That's the frame I'm carrying out of Shanghai and into the rest of this trip: not AI as spectacle, but AI as infrastructure — arriving in pharmacies, phones, forests and coffee orders alike, well before it ever reaches a boardroom. What happens next depends less on the models than on the institutions built around them.