China’s AI and fintech scene are moving fast, but not always as expected by outsiders. At AFAC (Advanced FinTech AI Competition) this year, three experts with deep industry insights outlined a practical, down-to-earth roadmap for how AI and fintech innovations in China can serve as a reference for global markets: treating AI as infrastructure, designing products to be “AI-native,” and planning for compliance and local distribution long before launch. AFAC, co-hosted by Ant Group together with leading universities such as Peking University, Fudan University, the University of Hong Kong, Nanyang Technological University, as well as industry partners including NVIDIA, has quickly become a showcase for early-stage products that are already moving from demo to deployment.
“AI is not a bubble,” said Eelee Lua, a Singapore-based venture building leader specialised in FinTech and RegTech, who judged this year’s start-up tracks and spends much of her time advising start-ups on market entry. “I think any business leader, including founders, should take this very seriously, because AI will fundamentally reshape the paradigm for success. Especially if your company is larger, the challenges you face will be more significant.”
Although technical gaps are narrowing thanks to global talent mobility, she believes strategic communication and market persuasion remain key differentiators. That perspective also echoes the design of AFAC competition, which in its startup track places emphasis not only on technical strength but also on how projects are presented, incubated, and developed. By combining a talent program with an early-stage incubation pathway, AFAC highlights that technology alone is not enough – teams must also learn to communicate their value and connect with markets.
Lua also highlighted regulatory alignment as a major factor in the global adoption of AI solutions, especially in fintech. “Compliance isn’t just a checkbox, it’s the foundation of entering any new market,” she noted. She shared an example of an Australian fintech firm that underestimated the time and cost of acquiring a license in Singapore, eventually seeing expenses double. “You need experienced local compliance leadership from day one, even if you’re still months away from generating revenue.”
This focus on navigating complex regulatory environments is reflected in the real-world challenges posed to innovators at competitions like AFAC. The startup track of this year’s AFAC, for instance, features themes like “promoting inclusive finance,” which requires participants to design solutions that are not only technologically innovative but also adaptable to diverse regulatory landscapes and local financial ecosystems. This alignment between competition challenges and industry pain points is what, in the view of many observers, gives AFAC its distinctive edge. Boxin Shi, a tenured associate professor in computer science at Peking University, emphasized that the competition’s strong roots in both AI and finance brings a strong practical focus. “AFAC is built around genuine industrial needs,” he said. “It features carefully designed tracks, professional scoring, and live demonstrations which all under a strict timeframe. This structure offers a standardized platform that brings together real-world problems, talented participants, and expert judges.”
This regulatory challenge, Lua suggested, is where cross-border collaboration and humility become essential – a mindset increasingly crucial for the next generation of fintech builders.
Jerry Yang, a venture investor focused on Africa and the Middle East, sees strong potential for “AI-native” startups-companies built from the ground up with AI integrated into their products, rather than older firms adding AI as an afterthought. At AFAC, he observed that tech startups in China have fully embraced AI, saying it’s no longer optional but a foundational layer. He sees this embodied in the competition’s startup track, which encourages proposals around “enhancing productivity in financial services” and “advancing financial data utilization” – areas where native AI integration can yield transformative gains.
He is particularly optimistic about AI’s applicability in emerging markets, especially in Africa, where he believes misconceptions about readiness abound. “Many companies in China assume African markets aren’t prepared for AI, often due to lower GDP and perceived willingness to pay,” Yang noted. Yet he emphasized that African businesses are eager to use AI to reduce development costs and accelerate digital transformation. Solutions that enable efficient, low-cost product development – such as AI-assisted programming and design – are in demand. The barrier, he explained, is not desire but accessibility: “They know solutions like Alibaba Cloud are cost-effective, but they often lack a direct channel to use them.”
This experience, Yang says, reflects a broader structural gap. “Many overseas genuinely want to cooperate with Chinese tech builders, but lack a proper channel. They can’t find one.” He emphasized that this isn’t a problem of interest but of accessibility. The absence of localized materials and support systems creates a barrier to adoption. “Many developers I know overseas say they sometimes can’t find good training materials or documentation in English or French that would allow them to use these technologies independently.”
That gap is partly why events like AFAC are drawing attention. Beyond showcasing solutions, the program has become a meeting point where early-stage founders can connect with mentors, investors, and industry partners.
Yang contrasted this with the approach of globalized tech giants. “Companies like Google or Facebook annually send large numbers of technical personnel to emerging market countries to train local developers. We don’t have such a mechanism.” He suggested that companies in China could significantly accelerate global adoption by investing in similar outreach programs, providing localized documentation, and establishing clearer pathways for engagement with global developers who are eager to leverage their AI solutions but lack the linguistic or cultural access points.
One concrete example of AI application comes from Li Xing, CTO of Kingdee Credit. The company first joined AFAC in 2024, where it won recognition for its data-driven lending tools, and returned to the 2025 competition with a new technical solution. As a two-time participant and award-winner, Kingdee has used the platform to showcase how its approach has evolved from data infrastructure toward AI-powered financial services. His team has built a novel financial AI system empowered by a massive knowledge graph from three decades of Kingdee’s enterprise data, covering more than 7.4 million small and medium-sized enterprises (SMEs).
The model aims to tackle persistent problems in lending: high risks and costs associated with serving SMEs, which often lack extensive credit histories. By mapping complex transactional relationships across supply chains – identifying who supplies whom, in which industry, and at which tier – the knowledge graph provides structural context that reduces AI hallucination and improves accuracy. Li detailed how their knowledge graph addresses the fundamental challenge of AI hallucination in financial contexts. “AI hallucination is essentially a problem that comes from the model’s architecture itself,” he explained. Large language models generate outputs based on probabilities from their training data, which can lead to inaccurate results. Their solution involves providing the model with “more relatively precise language materials and knowledge points” during operation, guiding it to solve problems in an organized manner according to established knowledge structures. This approach significantly reduces hallucination issues in financial applications where accuracy is critical.
“Banks often struggle to assess SMEs because their data is fragmented or absent,” Li explained. “We help financial institutions lend at the right time, with clearer risk insight.”
The system has already been adopted by more than 200 financial institutions in China, including major banks and online lenders. Now, Kingdee is exploring global market, with initial interest in Southeast Asia and Latin America. Li acknowledges that while some data patterns – like agricultural seasonality – may be transferable, local adaptation and partnership with regional financial institutions will be essential.
Across all three experts, a common theme emerged regarding the importance of understanding local market conditions. As events like Ant’s AFAC strive to create more interconnected innovation ecosystems, the focus is shifting from pure technological competition to collaborative problem-solving. The future may depend more on whether AI solutions can integrate smoothly into the fabric of global commerce – especially in places long overlooked by traditional fintech. As Lua put it, “It’s not about hype. It’s about reshaping how success is built.”
