China's AI Coding Agents Do A Lot More Than Just Code

Since 2024, agentic artificial intelligence (AI) has increased in popularity. However, while generative AI models like ChatGPT can create content based on user input and existing data, AI agents can plan, research, analyze, and even collaborate with little outside influence from an actual human. One of the central purposes of this technology is to aid developers in the production of better programs at lower costs — and it seems to be working. While generative AI has already impacted business practices in many industries, agentic AI can now complete much of the work in certain branches of major companies. However, this might not spell good news for everyone down the road as some reports suggest that the costs associated with using AI agents are significantly lower than the cost of paying actual employees.

While American AI companies like Anthropic and Replit have been dominant forces in the agentic AI industry, Chinese firms like DeepSeek and Zhipu AI are catching up. Speaking with CNBC, representatives from Zhipu revealed that the majority of the people using its agentic AI model were based in China and the U.S. Part of the reason for this popularity is the fact that, while most American AI agents are privatized and require significant investments to access, many Chinese AI agents are open source — meaning they're available to anyone. These effectively free programs significantly lower the barrier for entry and can offer any company seeking to cut costs, speed up product development, and reduce the size of their workforce a massive incentive to opt for Chinese AI tech instead of domestic alternatives.

According to Capgemini, 23% of organizations were already utilizing AI agents in 2025, with many companies relying on it to automate large-scale tasks in their IT, customer service, and financial monitoring departments. Agentic AI is also heavily used in software development, with A16z revealing that the release of new apps increased by 60% between December 2024 and December 2025.

Of course, agentic AI has yet to become a silver bullet for corporations: Major companies like IBM are already learning hard lessons about replacing workers with AI, and the technology can cost companies around twice as much as generative AI to implement into an existing workflow. Those costs can skyrocket depending on the complexity of the agent needed. CX Today estimates that the overhead costs of a simple AI agent can range from around $10,000 to $50,000, while a network of collaborative agents could cost companies millions. While the ultra-rich might have billions to invest in AI startups, those costs could prove seriously prohibitive for others. 

While opting for open-source agents can mitigate some of these expenses, the programs aren't exactly flawless. When experts at CNBC assigned a Zhipu AI agent and American AI agent the same task, Zhipu was able to complete its assignment faster, but the finished product was deemed less polished.