Editor’s note: This article was written by Lucia, a TechNode reporter.
For businesses adopting AI, the question is no longer simply which model to use. It is also where AI should run, how much it will cost, and whether sensitive data can remain under the company’s control.
PeakVision AI’s answer is to bring more AI work inside the enterprise. The system treats the task—not the individual app—as the basic unit of work: users state a goal, and AI agents handle the steps across a local computing environment.
From Research to an AI-First Enterprise Stack
Founded in 2024, PeakVision AI began with two obstacles to enterprise AI adoption: the difficulty of deploying the technology and concerns over data security, co-founder and CTO Liangyu Nie told TechNode.
In March 2026, the company established its Hangzhou headquarters and received RMB 15 million in government R&D support. It delivered its first enterprise systems and closed an initial funding round in May, followed by a second round worth tens of millions of yuan in August.
PeakVision AI is building an enterprise AI stack that combines locally deployed computing, FengKan AIOS, and AI-agent services. The system is designed around how agents complete tasks, rather than how people move among individual apps.
“You don’t need to download different apps. You just tell AI what you want in a chat box, and AI does it for you,” Nie said.
The startup’s central product thesis is that: enterprise software may increasingly be organized around intent and execution, not menus, files, and separate applications.
The company plans to open-source FengKan AIOS by the end of October.
Scaling Enterprise AI Through Hardware, Services, and Partnerships
PeakVision AI makes money from hardware, subscriptions, and customized enterprise AI deployments. Its wider goal is to turn project-specific work into reusable solutions for different industries.
FengKan AIOS supports third-party hardware, while PeakVision AI is testing a selective approach to its Skill ecosystem. Nie said end-user Skills, which may contain proprietary workflows, will remain untouched. More generalizable Skills built by partners and forward-deployed engineers could be released to the market.
The company is recently testing the model through an FDE academy launched by Shanghai Sixunited Intelligent Technology with Advanced Micro Devices (AMD)’s support. AMD provides computing equipment, technical toolchains, and project resources.
PeakVision AI also works with chipmakers early in product development to optimize performance and costs, including through a partnership with ChangXin Memory Technologies to reduce memory costs.
Competing for Small and Medium-Sized Businesses and Going Global
While enterprise-focused AI companies such as Wonderful serve large, often regulated organizations with AI platforms and deployment teams, PeakVision AI is targeting small and medium-sized businesses.
Nie said these businesses face the same challenges—cost, technical complexity, and data security—with fewer resources to address them. PeakVision AI uses lower-cost hardware from China’s supply chain to turn each deployment into a repeatable solution for similar customers.
“I believe small and medium-sized businesses around the world face the same challenges,” Nie said. “They have the same pain points as large enterprises, but one additional problem: they simply cannot afford to spend as much.”
The company is pursuing overseas opportunities, with distributor interest in the U.S., Israel, and Saudi Arabia, and is also exploring Southeast Asia, Latin America, and Africa, where local AI may appeal to businesses with limited or unreliable internet access.
From Finance to Manufacturing
PeakVision AI points to manufacturing and financial analysis as early examples of its approach.
At a lifeboat manufacturer, the company said its AI agents shortened a product-design cycle from one month to one week. At a yacht maker, AI-generated designs shifted engineers’ work from drafting proposals to reviewing and approving them.
The company has also developed tools for anti-corruption authorities to analyze financial records and for police to trace funds linked to fraud, according to Nie.
Beyond Automation: What Remains Human?
PeakVision AI aims to reach 80 to 100 enterprises across 10 industries by 2027, with longer-term ambitions to make FengKan AIOS a widely adopted enterprise AI platform.
Nie believes AI cannot replace human judgment but acknowledges that it could reduce staffing needs. As AI takes over more tasks, workers may increasingly focus on reviewing results and making decisions.
“For an employer, tasks that once required at least five people could be handled by just one,” Nie told TechNode.