On 4 June 2025, Ant International announced its AI strategy with AI Platform for Fintechs. Chief Innovation Officer Jiangming Yang shares here more thoughts on the direction of AI at the company.
Jiangming Yang, Chief Innovation Officer of Ant International
Q1: What is A+ GenAI Cockpit? Who does it serve? What does it do for the client? What is your potential client pool?
A1: It is an AI-as-a-Service platform for financial services providers (e.g. fintechs, e-wallets, banks, super apps) to architect an agentic and ultimately AI-native financial services, by combining a fintech-specific toolbox, dynamic industry knowledge base with business-ready AI innovations.
It integrates automated workflows and task orchestration with a dynamic enterprise context across main fintech tasks, from payment orchestration, customer onboarding, compliance checks to fraud detection, dispute resolution, as well as evaluation and performance optimization.
Developers may choose from our broad range of plug-in fintech AI agents and applications built from our deep industry expertise, or they may choose to build their own agents and applications using the AI infrastructure and architectural capabilities.
Our endgame here is about transitioning from agentic commerce to what we call AI-native commerce/finance.
Q2: In your view, what are the biggest gaps in current LLMs when applied to fintech, and how do you address these gaps?
A2: Despite the rapid rise of LLMs, their adoption in complex sectors like financial services remains limited. FinAI is seen as the ‘holy grail’ of AI applications due to three key challenges: high-security risks (e.g., deepfakes, hallucinations), the need for fast, expert-level decisions in high-stakes scenarios, and the high cost of deploying enterprise-grade AI.
A competitive FinAI needs to complete tasks (like integrating payment methods and handling payment processing under different compliance requirements) not only competently but also with tangible efficiency and cost advantages. General LLMs fall short in these fragmented environments.
While many players focus on isolated AI tools or workflows, fintechs need a future-ready strategy that not only improves efficiency and cost but also enables them to build a full AI-powered customer journey—and eventually, their own AI-native services.
Q3: When you say it is an AI-as-a-Service Platform for Fintechs, how is it different from a “financial agent builder” or a “financial LLM model” or other peers? What is unique about your platform?
A3: AI is great at generating content or writing code, but does it autonomously take expert actions or execute real financial tasks? Not so much, because many current models lack the necessary business context and workflow integration.
For example, if a customer asks for a refund, these tools might show the refund policy, but they can’t actually initiate the refund, update the ledger, or even recalculate loyalty points.
Second, by nature, AI for financial services must meet the highest standards for trust and security. We are talking about compliance, fraud control, and now commonly-known challenges of model hallucination or bias.
Cockpit is purpose-built for financial services. It combines deep industry know-how, strong compliance safeguards and robust foundational AI technologies to help our partners develop powerful agents that handle tasks on their own with high reliability, precision and security in real automated workflows.
Take KYC (Know-Your-Customer) as an example. It’s a process that organizations use to verify the identity of customers to prevent fraud or money laundering. KYC requirements vary by market.
Agents powered by Cockpit are equipped with financial domain expertise and a compliance knowledge base that can be continuously optimized. They can understand and respond to these localized requirements and execute tasks accordingly.
Q4: Could you walk me through how fintechs in different stages of growth may transition to AI-native services using the Cockpit?
A4: Cockpit supports fintechs at any stage. Startups or SMEs can get started quickly with pre-built agents, while more established organizations can tailor agents and workflows.
In the long run, we believe AI will fundamentally redefine the way the economy and society run. An AI-native future is where products, services, and organizations are built with AI organically from the ground up, not as an add-on. Financial services must evolve accordingly.
Q5: Could you share some preliminary results of Cockpit’s efficiency, cost or security gains from current use cases?
A5: Antom Copilot (an industry-first AI agent that helps merchants implement, orchestrate and optimize payment methods):
- 90% faster integration
- 90% satisfaction with issue resolution
Customer services use case:
- 70–80% of service requests are handled directly by AI
- Quality of AI answers matches or exceeds human experts in 70%-80% of cases
Falcon Time-Series Transformer FX Model (a transformer architecture-based big data model that helps businesses reduce FX-related costs and risks)
- 47% of A&I’s current transactions are cross-border and involve foreign exchange
- The AI FX model already handles 63% of our applicable FX funds flow
Q6: What are the key measurements of a successful AI platform？
A6: Exceeds Human expertise: Any meaningful technological revolution must bring qualitative improvements in efficiency and scalability. We believe AI can do this, but only if it truly understands the industry and supersedes the totality of industry expertise.
Scales quality services to a broader user base: In the past, financial services often depended solely on human experts (e.g. financial advisors, insurance specialists). Such services are uneven and expensive. With AI, these services can now be scaled and made more accessible to a wider client base, especially SMEs and otherwise underserved consumers, with comparable or even higher quality.
We look at two basic criteria:
- Efficiency and scalability: The AI platform should help financial service providers increase operational efficiency and scale business.
- Security, reliability and compliancy: The AI platform should leverage trusted technologies to mitigate internal risks and ensure external security.
Q7: What would the security and risk mechanisms look like in AI-Native financial services?
A7: Externally, there should be 360-degree protection in place to guard against external threats based on a dynamic threat survey. They may include prompt injection attacks, deepfakes, data leakage, etc.
For instance, Alipay+ GenAI Cockpit features an anti-deepfake system powered by adversarial data, a large bad case library, and robotic testing, with the ability to evolve through real-time feedback. Joint defense mechanisms help prevent tampering and spoofing.
Internally, risk control mechanisms should ensure that AI outputs remain accurate and compliant.
Backed by our deep expertise in the fintech sector, Alipay+ GenAI Cockpit delivers industry-leading defense models and risk control capabilities tailored for financial services.
For example, the platform uses over 100 recognition models and 600,000 risk lexicons to reduce hallucinations and detect adversarial prompts or data leaks.
Organizations can also upload their own risk lexicons to enhance detection precision.
Q8: Where do you see AI’s role in fintech in 3 or 5 years? Where do you see Ant International in 3 years, 5 years in terms of technology?
A8: I believe AI is driving a real fintech revolution. It will bring transformative changes in how financial service providers serve customers and scale their business.
That is why we're investing deeply in AI. I'm confident that Ant International will become a leading platform provider in supporting such transformation. Our 3-5-year roadmap focuses on advancing AI’s role in fraud detection, personalized financial advice, and automated compliance, while cautiously navigating the path toward AGI.
Q9: What is your own outlook for the timeline of artificial general intelligence (AGI), and how would AGI shape global commerce in the endgame?
A9: I am more on the cautious side regarding AGI, particularly regarding the actual “intelligence” of current models. Today’s large language models built on transformer architectures might be reaching the limits of their current capabilities by extracting existing internet knowledge.
I believe further progress toward AGI requires breakthroughs that enable models to learn from the physical world, like human learning. Simply improving efficiency in language processing will not result in significant advancements.
As we advance AI technology, it is crucial to consider its broader impact and purpose. Achieving excellence in sectors like finance requires integrating social and ethical considerations within AI development.
The inclusive, sustainable, and fair economy and society we strive for is still evolving, which demands a shared vision of the values AGI should embody. The “G” in AGI signifies more than just technical prowess and it must reflect societal and ethical values as well.
Q10: Can you outline Ant International's AI strategy and how it supports your strategic growth objectives?
A10: We are focused on 3 key areas in our AI strategy:
- Large investments in security and trust AI to help our clients protect themselves and their users,
- Vertical expertise and fintech-specific knowledge banks, and
- Building our B2B AI products as an AI-as-a-Service (AIaaS) platform for fintechs.
Secure and trusted AI capabilities as a competitive edge: We invest heavily in security AI capabilities, including anti-deepfake tools and hallucination detection, to help partners serve large and small merchants across complicated markets.
Fintech-specific expertise for precise decision/action-in-context: Our AI solutions integrate workflow execution and industry-specific insights to empower our partners.
Antom Copilot, the world’s first AI agent for streamlined merchant payments, automates payment orchestration, dispute resolution, and conducts risk management via natural language, delivering 90% faster integration and achieving 90% issue-resolution satisfaction.
AlaaS Platform for fintechs: Alipay+ GenAI Cockpit gives fintechs, banks, e-wallets and super-apps a platform to build not only modular, custom-made AI agents and vertical solutions, but also develop a comprehensive suite of AI-powered fintech services. Partners can also develop AI-native services on this platform.
100% of Ant International’s TPV is AI-related: AI powers every step of the payment process, from risk and fraud prevention, payment success rate optimization to FX and liquidity management, delivering efficiency and scale to support our global growth.
