“The definition of a future company will be rewritten — a company will no longer be a registered legal entity and a collection of employees, but rather an agile living organism made up of creative individuals and a legion of AI agents,” said DeepWisdom founder and chief executive Wu Chenglin.

Wu’s vision underpins the rapid rise of his Beijing-based startup DeepWisdom, whose AI-agent platform Atoms aims to allow individuals with no technical background to build full-fledged digital businesses using natural-language instructions. The company is positioning itself at the intersection of generative AI, automation and entrepreneurship, betting that software development — long the domain of trained engineers — is about to become a mass-market activity.

“If you can describe your idea, you can build it,” Wu said in an interview, arguing that AI agents can replace much of the technical labor traditionally required to launch a business. “What we are creating is not just a faster coding tool. It’s a way to make entrepreneurship accessible to anyone.”

Wu, 36, previously worked at Huawei and later at Tencent, where he led teams building large-scale AI systems for search, recommendation, user profiling and natural-language processing used by hundreds of millions of people. He became one of Tencent’s youngest senior technical staff before leaving to pursue entrepreneurship.

He said he had long kept a notebook of business ideas, eventually deciding that the biggest opportunity lay in lowering the barriers between people and technology.

“There is still a huge gap between what AI can do and what ordinary people can use,” Wu said. “Our goal is to allow anyone to command a team of intelligent agents as if they were managing a professional company.”

That idea became DeepWisdom, founded in 2024, and its flagship product MetaGPT — later rebranded as Atoms.

Atoms allows users to describe what they want — a website, an app, a marketplace, a game — and then orchestrates a group of AI agents to perform market research, product design, engineering, deployment and operations.

A user wanting to build an online jewelry store, for example, can instruct Atoms to research competitors, identify a niche, design the interface, write the code, integrate payments, deploy the site and even analyze early user data.

“Users don’t hire a programmer,” Wu said. “They hire a software factory.”

DeepWisdom says the platform is used by small business owners, freelancers, designers and product managers who want to test ideas quickly without hiring technical staff or raising significant capital.

The company says a product manager in Canada used Atoms to launch a home-design website and gain paying users within a month. In South Africa, rescue workers built a calculation tool that reduced field computations from minutes to seconds. In the United States, an animation studio uses Atoms internally to speed up creative workflows.

Wu said the most meaningful case for him was an elderly man in Canada who built an educational product for his grandchildren. “It showed us this is not just about efficiency, but about human connection,” he said.

DeepWisdom launched publicly in February. Within a month, it reached $1 million in annual recurring revenue without paid marketing, according to the company, driven largely by community adoption and open-source exposure.

It now has more than 500,000 registered users globally and has ranked at the top of Product Hunt’s global charts multiple times, the company says.

The startup has raised hundreds of millions of yuan from investors including Ant Group, Baidu Ventures, Cathay Capital and Concept Capital, making it one of the most heavily funded Chinese startups in the AI-agent segment.

DeepWisdom employs around 80 people and plans to grow to 100-120 by the end of 2025, Wu said.

The market for AI coding and automation tools has become crowded, with companies such as OpenAI, Google, Anthropic and numerous startups offering tools to generate code, automate tasks and assist developers.

Wu argues that Atoms’ differentiation lies in moving from “vibe coding” — generating snippets of software — to “vibe business,” in which AI agents manage the entire lifecycle of a product.

Atoms includes built-in login systems, databases, payment integration, deployment tools and analytics, allowing users to launch revenue-generating products within minutes, Wu said.

A key feature is “Race Mode,” in which multiple AI models such as GPT, Claude, DeepSeek and Qwen work on the same task simultaneously. Their outputs are evaluated across dimensions including performance, security and maintainability, and the system selects or combines the best result.

“This allows users to act like a boss reviewing multiple teams,” Wu said.

Wu said security and privacy were designed into the platform from the start. Atoms integrates user authentication, encrypted data storage, secure payments and privacy protection by default, the company said.

DeepWisdom has integrated with payment platforms such as Stripe and says it complies with regulations including the European Union’s General Data Protection Regulation. It has also obtained SOC 2 certification, Wu said.

Beyond commercial products, DeepWisdom invests heavily in open source through its Foundation Agents initiative, which has attracted more than 150,000 stars on GitHub.

Wu said open source is critical to setting standards for agent protocols and accelerating innovation. “It’s not just altruism,” he said. “It helps build an ecosystem, attract talent and establish leadership.”

Wu’s long-term vision is what he calls the “one-person company” — a model in which individuals use AI agents to perform functions once requiring entire organizations.

In this model, entrepreneurs focus on ideas, strategy and creativity, while AI handles execution.

“This is decentralization and automation combined,” Wu said. “It changes who can participate in the economy.”

He argues that this will allow individuals to serve niche and long-tail markets that large corporations ignore, potentially reshaping innovation and competition.

“It’s not just a change in tools,” Wu said. “It’s a change in production relationships.”

Analysts say such platforms could reduce startup costs, accelerate innovation and increase competition — but also raise questions about labor displacement, data control and accountability.

“If anyone can build a company, the definition of a firm becomes fuzzy,” said one venture investor who has reviewed DeepWisdom’s product. “That’s powerful, but also disruptive.”

Wu acknowledges the risks but says the trend is inevitable.

“Technology always lowers thresholds,” he said. “This is simply the next step.”

As AI systems become more capable, Wu believes that economic power will shift from institutions to individuals — from organizations built on hierarchy to networks built on ideas.

“The company of the future,” he said, “will not be a building, a brand or a headcount. It will be a living system — and anyone can own one.”