Fabarta leverages AI to unlock value for firms

China is doubling down on artificial intelligence as a pillar of industrial modernization, spurring AI startups like Fabarta to leverage data-driven technologies to help firms unlock the value of data and sharpen competitiveness.

The State Council has unveiled its most detailed road map yet to deeply implement the "AI Plus "initiative. It calls for priority and stronger infrastructure to integrate AI across sectors from chemicals to finance.

Gao Xuefeng, founder and CEO of Beijing-based AI company Fabarta, said: "While interest in AI is at a peak, most enterprises are failing to capitalize on their own information assets."

He made the remarks on the sidelines of the China International Fair for Trade in Services that concluded in Beijing last week.

"Enterprises feel unprecedented urgency to deploy AI or risk losing competitiveness, but fragmented data remain the biggest bottleneck," he said.

To address such bottlenecks, the company has developed a cloud-edge-terminal architecture that helps firms make practical use of the vast data they already generate but rarely analyze.

Gao said that it has already built more than 30 industrial AI applications for over 30 State-owned and leading companies in sectors including chemicals, manufacturing, energy and finance.

One platform, developed with Sinochem, integrates knowledge engines and domain-specific models to improve document recognition accuracy to 99.2 percent and cut response times for digital operations manuals to under half a second, he added.

The approach reflects Beijing's broader aim of shifting from consumer-facing applications of AI toward embedding it deeper in production, logistics and enterprise services.

Industry experts argue that generative AI, or AIGC, can underpin "intelligent agents" that transform how industries operate, provided data bottlenecks can be resolved.

Fabarta is promoting what it calls a three-step strategy: co-developing high-value use cases with industry leaders; replicating proven applications for smaller firms through cloud deployment; and eventually bringing lightweight personal AI agents to consumers.

That strategy, according to Ge Shuang, co-founder and chief operating officer of Fabarta, is designed to "make AI accessible to companies of all sizes".

Ge said: "Large State-owned firms want platforms that can handle multiple scenarios, such as predictive maintenance tools for factory workers.

"But small and medium-sized enterprises often lack resources. For them, lightweight AI agents that replicate the capabilities of industry leaders can deliver fast, cost-effective results," she noted.

To address that demand, Fabarta has rolled out three tiers of its "cloud-edge-terminal" service, tailored to different scales of enterprise operations. The company said the model has helped it triple revenue growth this year.

Fabarta is expanding its reach abroad. Through a subsidiary, it has built cross-border AI agents that handle Customs clearance, tax refunds and compliance tasks.

In partnership with ByteDance, it has developed multilingual processing tools that it says improve efficiency in handling Customs documents by 60 percent.

The company is not limiting itself to enterprises. In July, it launched a personal AI assistant for writing, translation and data analysis. The product runs primarily on local devices but can tap cloud resources when necessary, a hybrid design that Gao said balances efficiency with privacy.

"AI is not here to replace people, but to free them," Gao said. "Our goal is for every individual, every organization and every industry to have its own intelligent partner."

Gao acknowledged that challenges still remain, but argued that a "data-centered" approach is the best path forward. "Models will continue to evolve, but without high-quality data and industry-specific knowledge, they remain generic," he said.

For Fabarta, that means embedding itself into the full value chain of clients, from research and production to supply and after-sales service.