Suppose there is a job like this: 60,000+ yuan a month, no deep tech required, no elite degree filter—hit the boom and you can turn your life around. Would you take the bet?
On September 16, Maimai released data showing that in the first half of 2026, the average monthly salary of newly posted AI roles broke past 60,000 yuan. Sales roles rose the most, up 27.8%.
Over the past year, the AI industry has raced into commercialization and spun out a wave of new jobs. AI sales, with relatively low barriers and a high ceiling, has become—in many small-town young people’s eyes—the closest path to success besides being a “test-taking machine.”
Is AI really that valuable? What is life like for most people in the field?
We talked to several front-line AI salespeople and got a sobering answer: only a few stand at the tip of the pyramid. More grassroots AI sellers take a base pay of a few thousand yuan, push immature products, and feel their way through an industry that has barely taken shape.
How do we cross the dark toward the light? The answer is firm belief—even faith.
Today’s story is about AI, but beyond compute it also carries the dreams of young strivers.
From foot-bath receptionist to selling AI
Zhou Jin, born in 2005, sees himself as part of a generation whose youth was threaded through with AI.
But he truly felt AI’s pull at a relative’s foot-bath shop.
Summers in Guangdong arrive early, humid enough to leave you gasping. The summer after graduation, Zhou stood at the front desk as a receptionist.
One day a young man about his age walked in, pitching an AI operations system.
In under ten minutes the manager called the shop manager, who then called the owner—Zhou’s fourth uncle.
The salesman first demoed a complicated flow on a laptop, hammered phrases like “AI automation,” “dedicated guidance,” and “lifetime support,” then put a metal box on the table.
Uncle, always shouting about “keeping up with the AI era,” did not hesitate—he cheerfully paid 30,000 yuan.
That was the first time Zhou felt how close AI was to money.
Only after joining the trade did he learn that AI sales sits on fundamentally different tiers.
At the top: sell compute and large models to big enterprises and government—no credentials or connections, no entry. In the middle: custom AI solutions that filter for degrees and experience. At the bottom: sell assorted AI SaaS tools and companion AI hardware.
For Zhou, a junior-college graduate, the first two paths were too narrow and too competitive. Selling AI hardware was the only boom-adjacent door he could reach.
People used to say small-town youth had only one way out: exams. Now the AI wave has finally given those who do not excel at exams another road.
But when a road is too easy, that is not always good. As Lu Yao wrote, spring roads are full of mud.
Three days to get in
AI hardware, as a niche inside AI sales, leans more toward sales than toward AI.
That nature puts it a bit farther from tech and a bit closer to playbooks.
Several front-line sellers mentioned the same point: money is hard because the product is hard to sell.
“Honestly, most brick-and-mortar shops do not have a real need for AI. What measures us as salespeople is whether we can create demand.”
In a fourth-tier city near Shanghai and Hangzhou, Du Sheng has done AI sales for almost two years.
When he started, the company gave him only three days to adapt.
Day one: a full day of pep talks—”over and over about the AI boom and the company’s future, nothing solid.”
Day two: nearly 60 hours of videos, expected to be self-studied in a day.
Day three: a test; five of ten were cut.
The five who stayed already had some foundation. “Recruiting says zero barrier, but if you are a true beginner, you will be washed out.”
For all the dramatic onboarding, once on the job he found the bar for AI hardware sales was low: “You do not need deep AI knowledge—memorize the PPT talking points and explain the demo video clearly.”
He concludes that this rough training matches the customer base.
Most clients are shop owners who “do not understand AI, buy the trend, but lack strong willingness or ability to spend.”
So an invisible technology rarely opens wallets—it has to be packaged as hardware to sell.
And to make buyers feel the spend is “worth it,” the pitch gets packaged too.
The most useful move: inflate product credentials.
When Du visits shops, he often uses company lines like “led by big-tech executives, a 20-person Tsinghua–Peking research team…”
In reality he has never heard of those people in any real setting. When clients report issues, the company’s single programmer handles them.
Next: hard promises.
“For example we propose a bet: guarantee 1:5 ROI or refund.”
Those stay oral, then contracts quietly redefine terms—say a client pays 10,000 yuan in fees, and as long as tracked revenue hits 50,000 yuan, the goal is “met.”
Du saw the risk soon after joining but did not leave: “Before this job I had been unemployed for three months and lied to my parents that work was ‘going well.’ I could not drag it out anymore.”
After joining he did earn decent money—over 10,000 yuan in good months—but at a cost.
“Our product does have some use, but even so, for every shop I close I mark it on Amap and never dare go back.”
An AI that does not exist
At least Du is selling something that actually exists.
More extreme: some young people who entered chasing an AI dream later find they walked into the wrong door entirely.
Zhou never expected that behind the first “AI hardware” he sold, half the work was done by real people.
“When I applied and interviewed, the company looked normal—the role said AI store digitalization and offline customer-acquisition systems. Only after onboarding did things feel off.”
The company was tiny: four people including him—two support staff, two sales.
What Zhou sold was a so-called “AI customer-acquisition robot.”
The workflow: support staff register female accounts on dating apps, then an AI posing as a beautiful woman chats up nearby users.
“It initiates steamy talk and sends photos. Photos scraped from overseas sites; scripts prepared in advance.”
At key moments the backend flags a human to take over, suggest meeting offline at local internet cafes, pool halls, and similar venues—whose owners are the company’s clients.
Owners know the crude program will not sell for much, so they wrap it twice:
First, appearance: stuff the software into a second-hand mini industrial PC and a custom plastic shell. Sold as hardware, unit cost under 200 yuan.
Second, pitch: Zhou must call it an “AI smart butler” and must not explain the real mechanism—package it as a complex AI ops system.
“The company finds demo videos of professional systems online for us to show clients. Most people cannot tell; emphasize ‘customer acquisition’ and you can usually get by.”
Zhou quit after a month. Base pay 1,500 yuan; devices rented monthly at 399 yuan a set; he took only 30 yuan commission per deal. Little money, daily lying to clients—his conscience could not take it.
Shabby crews and gray businesses are still edge cases. Du told me insiders know the core problem with AI hardware: customers have no real privacy protection.
Privacy? Does not exist
One of Du’s company’s AI products helps shops run online ads and business analytics.
For the AI to work, clients must open backends and sync large amounts of operating data—orders, cash flow, campaign results, even revenue—all visible to the system.
Inside the company those data were not strictly isolated: “To demo results easily, the company gave us eight sales backend accounts.”
The subtext: when needed, to convince clients AI can truly “empower” brick-and-mortar, sales are quietly allowed to show partner clients’ private data.
Last month Du closed a barbershop—a chain spanning five provinces.
Two days later, pitching the next client, he casually opened that shop’s backend.
“I know it is wrong. Every time I say ‘just look, don’t spread it.’ But without real cases, who believes you?”
A colleague told him something worse. A client using an AI product with Q&A forgot it was a store system and treated it like Doubao, dumping everything into it.
Illnesses, debt disputes, even private relationship matters all showed up in the backend—and became office gossip.
One day the colleague said he would stop talking about that client: the person had uploaded a half-nude photo that was hard to look at.
That may be the truly rough edge of bottom-tier AI today: beyond immature tech, the worst outcome is that there are no rules at all.
When small firms rush to monetize and sales carry quota pressure, many boundaries go unsupervised—left to conscience alone.
A foreseeable future
These messes may be growing pains for sinking AI commerce in early commercialization, but they should not make us reject the whole direction.
Because the people inside are far more certain about AI’s future than bystanders.
That belief starts upstream with the big tech firms.
Elson, who once did AI research at Tencent, said Tencent earlier sorted AI-company roles into five types: enablers (models), collaborators (product and delivery), promoters (sales and marketing), governors (security and compliance), and supporters (ops and logistics).
Among them, “promoters” rose 8 percentage points over the past year—the fastest of any category.
“In 2024 everyone competed on R&D; by 2025 products matured and the whole industry sped up commercialization.”
From a big-tech view, tech needs to land in sinking markets. Local agents and small founding teams also want to hitch a ride and help pave the road.
Du’s company is a microcosm. Its main business is e-commerce; the past two years it has largely used e-commerce profits to fund AI, and the biggest gain is a local base of 2,000+ merchants.
“The company plans to win regional agency rights for a major model vendor. Once we get it, the money will come back eventually.”
Look further down: are all the shop owners who pay simply being scammed? Not entirely.
Closest to the sinking market, Du and Zhou share a sense: ordinary people’s acceptance of AI is higher than expected.
Du still remembers one owner: “I watch AI short dramas, find AI customer service online, search with AI. If this is the future, why not try it in business?”
He thinks most AI hardware on the market has limited real effect; owners who buy usually have the readiness and ability to take risk.
What connects big-tech ambition to owners’ belief is countless front-line sellers like Zhou and Du.
Last year Zhou left home for Hangzhou. To him, Hangzhou is the best “training ground” for AI sales—opportunity everywhere, competition fierce.
His base is now 5,500 yuan plus 20% commission. Money is still thin, but at least he touches real products, clients, and a full commercial chain.
“In the real-estate era, good salespeople did not obsess over whether a project was 10,000 or 100,000 yuan per square meter. I think the AI era is the same. I believe I am on the boom—keep walking and there will be a chance to break through.”
Sometimes an ideal picture does not need absolute correctness to pave the way; as long as enough people buy in, the future arrives faster.
It is these uneven beliefs that push AI from big tech to street corners—with the most solid shove from behind.
相关文章
