JD.com puts 60 AI tools in merchants' hands; commercial property still runs month-end billing in Excel

On September 23, JD.com folded more than 60 AI tools into its Jingmai AI Business Centre. Merchants can now ask questions in natural language and let the AI do the work. JD.com says merchant activity on Jingmai AI tools rose more than 150% over the past three months.

The same day, we thought of a different scene.

Month-end at a wholesale market. Two people reading meters across dozens of stalls, entering the numbers into Excel line by line.

That gap is what this piece is about.

On the e-commerce side, AI is already wired into daily work

On September 23, 2026, JD.com held a briefing in Beijing on its Jingmai AI Business Centre, consolidating five modules: a super assistant, AI experts, full-chain AI diagnostics, a third-party tool marketplace and merchant-built tools. It also launched its first "Jingmai AI Expert Team." Merchants ask in natural language and the system handles batch processing, automated inspection and intelligent setup.

A Jingmai AI executive told Cailian Press that the shift is from "giving tools" to "giving results," with all operations requiring merchant authorisation and every task traceable.

What is worth noting is not the number of tools but the name: "business centre."

E-commerce has pressed AI onto the tasks merchants actually do every day: reading data, diagnosing problems, taking action.

And on our side?

Month-end settlement is the least visible and the hardest to avoid

A Phoenix News report on Foshan's commercial property stock era laid out an old industry ailment plainly: pre-planning firms deliver proposals without implementation, leasing teams sign tenants without managing operations, and operations teams inherit a positioning that no longer matches the market — falling into a cycle of "revise and revise again, lease and vacate again."

Leasing is the visible difficulty. So resources, tools and dashboards all pile up at that end.

What nobody films, and nobody writes case studies about, is month-end.

Meter reading, cost allocation, invoicing, chasing payment.

Water and electricity meters are read stall by stall. Common energy consumption is allocated by area or by contract terms. Rent, property fees, utilities, late charges and waiver clauses all differ. Once the bills are out, dozens or hundreds of tenants get phone calls and WeChat messages, one by one.

This happens every single month. Done well, it is just the job. Done wrong, it is an argument with a tenant.

It never appears in a leasing report, but it consumes fixed hours from operations and finance.

The data is already there; it just never reaches the invoice

A smart parking solution for commercial complexes published by Jieshun Technology offers an idea worth a look for property and market operators: the system does not just manage vehicle entry and exit. It has open integration capability and can connect with a complex's membership and consumption systems, supporting automatic parking-fee waivers when spending thresholds are met, member point deductions and reserved parking. Operators can also use parking data to see customer visit patterns and dwell times.

Parking was once recorded by security guards. Now it is part of membership operations.

What about electricity meters? Water meters?

Most commercial complexes and wholesale markets installed smart meters years ago. The readings are digital and upload automatically. But the person walking the meters at month-end is still there.

The data sits in one system, the invoice in another, and a human moves it across.

So when operators ask "what AI should we actually adopt," the answer is often not that data is missing — it is that the data never lands on the same invoice.

Do the operating maths first, then talk about AI

A report on AI implementation by East Money put it well: as general model capability becomes foundational, what enterprises are truly willing to pay for is not "having an AI" but whether AI can embed into existing production processes and change efficiency, cost or revenue. That is the threshold where enterprise AI investment moves from a "technology budget" to an "operating budget."

In commercial real estate, the maths is straightforward.

Investing in AI on the leasing side carries uncertain returns — one more or one fewer signed tenant involves too many variables. Investing in AI at month-end carries certain returns: a fixed monthly consumption of labour, a known cost in hours when an error requires explaining to tenants, and a measurable delay in cash collection for every day billing slips.

That is why, when looking at a leasing operations platform such as SHEYU AIMARKET, we suggest starting with the billing engine rather than the leasing dashboard.

How to implement it

SHEYU AIMARKET is a leasing operations platform for wholesale markets and commercial complexes. It puts asset and lease control, leasing CRM, a contract billing engine, property IoT, internal office workflows and food-supply-chain delivery into one system, with an AI steward handling leasing follow-up, bill chasing and meter-to-invoice processing.

For the specific problem discussed here — monthly settlement — it takes over two steps. First, on the property IoT side, meter readings feed directly into the invoice, with no manual transfer. Second, the contract billing engine plus AI steward issues invoices according to contract rules, sending and chasing as required.

The repetitive work from meter reading to payment chasing runs by rule in the system, and people step out of it.

No exaggeration: it will not close a tenant for you, nor will it fix a wrong market positioning. It handles the thing that happens every month without fail and that nobody wants to do.

For commercial real estate, the first place AI should land may not be the leasing dashboard. It may be the month-end invoice.