SHENZHEN – A revolution of kinds is going down at a Guangdong software program growth agency.
In simply three months, Guangdong Create Science and Technology has produced some 50 software program methods, up from a mean of 1 or two per 12 months, stated Mr Huang Qijun, the agency’s chief AI (synthetic intelligence) officer.
This is all due to a military of AI brokers, designed to autonomously replicate what the agency’s software program engineers do at every stage of the event course of.
Rolled out in May, these brokers can do in a matter of hours or days what would have taken people months. “And you don’t have to pay them a salary,” stated Mr Huang.
The use of AI has given the agency a transparent edge over its rivals, he stated. The firm can now slash costs by some 70 per cent, and submit tender bids with a model of the requested software program already enclosed.
“This is a revolutionary thing,” he stated. “Why wouldn’t people use it?”
Mr Huang’s agency is one in every of a rising variety of companies making use of AI to their workstreams, as China ramps up efforts to diffuse a know-how seen as a method to turbocharge the economic system.
Some 53 per cent of 226 Chinese corporations say they’re integrating AI throughout a number of workflows – 11 proportion factors greater than the worldwide common, in line with a May 2025 survey by the consulting agency Accenture.
Still, the federal government has loftier targets.
It launched in late August a roadmap to speed up the combination of AI throughout China’s economic system and society, an initiative identified as “AI Plus”.
It desires the adoption of AI brokers and clever units to exceed 70 per cent by 2027, and 90 per cent by 2030. More superior than chatbots, AI brokers can automate advanced duties and take actions to realize targets with minimal human supervision.
Industry gamers and analysts whom The Straits Times spoke with say that China’s adoption of agentic AI has risen steadily this 12 months.
The nation’s market for AI brokers is anticipated to quadruple from 2024 to succeed in 10.9 billion yuan (S$1.96 billion) in 2025, in line with an August projection by Hap Academy, a Beijing-based tech trade think-tank.
Chengdu Minto Technology, a developer of AI brokers, has seen demand for its merchandise rise by 30 to 40 per cent in 2025, says deputy basic supervisor Zhou Bo.
“Now many organisations are actively embracing AI, and are starting with small applications and use-cases,” he stated, including that the majority of his prospects had been authorities our bodies and companies.
A Suzhou-based gasoline firm, for instance, had engaged Mr Zhou’s agency to develop brokers that might traverse its inside methods to supply standing updates on how initiatives had been progressing, and spotlight choke-points if any.
Some authorities businesses, he stated, had been additionally tapping brokers that put collectively experiences by analysing information from a number of bureaus.
To additional increase gross sales, Mr Zhou’s firm packages a few of its AI agent software program along with {hardware} – as Chinese patrons are usually extra prepared to pay for the latter than the previous, he stated. “If you buy a piece of software, you feel like you can’t really see anything. But if you buy hardware, it’s right there.”
One such providing is a water-bottle sized cylinder housing a digital Chinese fortune god, to which people can pose questions. Its price ticket of 1,999 yuan comes with a subscription to the corporate’s proprietary agentic AI software program, which might in any other case command only a few hundred yuan.
Chengdu Minto Technology’s cylinder which homes a digital Chinese fortune god.
ST PHOTO: JOYCE ZK LIM
Even as extra Chinese corporations heat to AI, attitudes to this nascent know-how fluctuate broadly, from sceptics to fervent adopters, says Mr Cao Kaibin, the president of Hap Academy, which surveyed about 100 firms this 12 months on their use of AI.
He estimated that about 30 per cent of Chinese companies had been experimenting with the use of AI, whereas about 5 per cent had been raring to use the know-how to get forward of their rivals.
Firms that invested probably the most in AI brokers had been within the finance, software program and web, and manufacturing industries – sectors the place their use circumstances are most clearly outlined, he stated.
Many of probably the most enthusiastic adopters of AI, he added, had been small corporations whose operations had been extra nimble, and had been “more willing to try it out even though there might be risks”.
At the Guangdong-based software program firm, for occasion, Mr Huang isn’t fussed about handing over the lion’s share of software-writing to AI brokers.
His engineers examine the software program earlier than submitting it to prospects, he instructed ST. “And at the end of the day, when the customers check it upon delivery, they will know whether it passes their requirements or not.”
Still, the “vast majority” of corporations, by Mr Cao’s estimate, undertake a wait-and-see perspective in the direction of AI. A key concern, he stated, was that the know-how continues to be not correct sufficient for firms to completely belief.
For now, only a small fraction of Chinese firms report sizeable good points from utilizing AI. In a report launched in July this 12 months, Accenture discovered that solely 9 per cent of the 163 Chinese firms it surveyed noticed vital progress in productiveness, income or income as a results of AI adoption.
Mr Meng Yin, founding father of Shenzhen The Nine’s Light Technology, a start-up which sells AI brokers, stated one problem which firms face when searching for to use the know-how is an absence of readability on what precisely they want.
As AI use rises in China, questions come up about what this can imply for an already weak job market. In July, China’s youth unemployment charge reached 17.8 per cent – its highest stage since August 2024.
Ms Wu Manhua, whose Guangzhou-based agency makes use of AI brokers to assist corporations analyse dangers and alternatives within the mental property house, says that experiences which as soon as might take as much as every week to jot down can now be carried out in a number of hours.
And whereas her agency continues to rent analysis and growth personnel, she reckons that its demand for administrative employees, together with customer support officers who write these experiences, will drop considerably.
“Normally, you’d need about 10 or 15 people to serve 100 companies,” she stated. “But with AI… perhaps one or two will do.”