NVIDIA AI Podcast, Episode 291
Title: AI Agents and the Future of Global Trade with Alibaba’s Kuo Zhang
Date: February 27, 2026
Host: Noah Kravitz
Guest: Kuo Zhang, President of Alibaba.com
Episode Overview
This episode dives into how AI—specifically Alibaba.com's new agentic platform, Axio—is transforming the world of global B2B trade. Kuo Zhang, President of Alibaba.com, shares firsthand insights on the development and deployment of AI agents, the challenges of global commerce, enabling small entrepreneurs, and the evolution of trust, automation, and scale in the age of intelligent systems. The conversation explores both technical aspects and the broader vision for how AI is lowering barriers to global trade—moving toward the goal of making cross-border business as simple as online shopping.
Key Discussion Points & Insights
1. The Evolution of Alibaba.com & The Role of Technology
Zhang outlines Alibaba.com’s journey:
- Founded in 1999, evolved from “yellow page” to the world’s leading B2B platform
- Connects 50 million buyers and 200,000 suppliers annually, enabling $60B in transactions
- Zhang’s own journey: Started at Alibaba in 2011, served on B2C (Taobao, Tmall) and B2B sides, became Alibaba.com President around 2021
- Dream: “To make global trade as easy as online shopping.” (03:10)
How technology changed B2B commerce:
- Built core platforms: Search/demand-supply systems, transaction/payment/logistics, trust solutions
- Enabled cross-border trust systems and streamlined operations
- Now entering the AI era for even greater impact
2. Introducing Axio: An AI Native Application for Global Trade
The Vision and Functionality
- Axio is built on state-of-the-art large language models
- Differences in user experience: Users now enter natural language requests describing full use cases and scenarios, rather than brief search terms (05:25–06:09)
- Example: “Now users can describe in a full sentence, like what is the scenario this battery is used in… It looks like a kind of suitcase, it’s portable …” — Kuo Zhang (06:10)
- Axio understands, parses, and breaks requests into actionable components to match products/suppliers
Shift from Search to Agentic Model
- Goes beyond search—users can assign complex, multi-step tasks and Axio can autonomously deliver results
- Expands reach: People new to global trade can participate easily; only ~30% of Axio users previously used Alibaba.com (07:36)
- “It’s completely lowering the barriers for people to enter into this field.” — Kuo Zhang (07:36)
3. Real-World Examples of Axio’s Automation
Examples of automated global trade tasks:
- Bolivian Games Sourcing Case (08:18–11:22)
- Supplier for a multi-sport event in Latin America uploads an Excel of hundreds/thousands of items
- Axio interprets requirements, applies compliance rules, finds suppliers, and completes a week(s)-long job in hours or minutes
- Automates communications and supplier matching
- Idea to Product Launch
- Example: Designing clothing for ADHD children
- Axio conducts market research, suggests product designs, recommends manufacturers, and supports prototyping
- Covers journey from concept to product to finding suppliers
Summary:
- Axio enables both sophisticated businesses and total novices to leverage global supply chains, automating nearly the full life cycle from ideation to execution (11:40–13:23)
4. Agentic Collaboration, Boundaries, and Human-AI Interaction
How Axio Orchestrates Tasks
- Accepts requests (text, drawings, files), decomposes into tasks, and executes them autonomously
- Involves users where needed for decision-making or iterations
- Uses iterative evaluation and integrates platform feedback to improve outputs
Boundaries & Human Oversight
- “…Whenever actually there’s a decision cannot be made by… the machines… it will involve the human to make the final decisions… Or when AI exceeds its boundary, it will come back to the humans … This is how we execute this whole system.” — Kuo Zhang (13:23–14:58)
- Iterates solutions where qualitative, nuanced, or high-stakes (e.g., negotiation, compliance) decisions are needed
5. Building AI for a Global and Culturally Diverse Audience
- Three layers for global readiness: (15:32–17:15)
- World and domain models: Local rules, compliance, and domain-specific knowledge baked into models
- Proprietary data: Over 200,000 suppliers, their product details, and certification data help create contextually relevant matches
- Human-machine hybrid: Where AI lacks, humans step in to ensure quality, safety, and compliance
- System is continuously improved by observing real-world user interactions and iterating on results
6. Capturing Local Nuance and Industry Knowledge
- Three-layer approach: (17:55–19:54)
- Dataset foundation: Leverages massive product, supplier, and transaction data to extract domain knowledge
- Industry know-how: Relies on expert systems to evaluate and improve AI outcomes
- Platform iteration: Deployment on Alibaba.com allows for measurement of conversion rates and outcomes, with systematic improvement based on feedback
7. Trust, Guardrails, and Building for SMEs & Entrepreneurs
- Human-in-the-loop safeguards for high-stakes or uncertain areas (21:30)
- Iterative machine learning optimized by platform KPIs: If results aren’t effective, they are retrained and updated
- Lowering barriers:
- Huge response from solo entrepreneurs: “Among these applications I think more than 40% of them mark them as a solo entrepreneur.” — Kuo Zhang (23:14)
- Axio supports product finding, design, and supplier scouting for novices
- Expands on Jack Ma’s original mission: “To make it easy to do business anywhere.”
8. Technical and Business-Model Surprises
- Technical: Qualitative, high-value B2B tasks pose new challenges vs. B2C commodity e-commerce
- “The questions become much more expensive… many times it’s not quantitative, it’s a qualitative task.” (25:22)
- Business model: Shift from keyword-based searching and sponsored listings to matching intent requires rethinking revenue streams (26:57)
9. Advice for Executives Scaling AI Products
- Start with the right question:
- “Whether your question is a real question or… a big enough question, I think that is the first one.” — Kuo Zhang (27:18)
- Three practice layers:
- AI-native applications (e.g., Axio): Rapid prototyping and quick iteration
- AI + Core Business: Apply AI to boost established businesses at scale (buyers and suppliers)
- AI KPI inside organization: Every role at Alibaba.com is measured on AI metrics, driving a pervasive culture of adoption (29:17)
10. The Next Decade: Envisioning the Global Trade Future with AI
- Impact on Global Economy:
- “The success of AI is whether or not we can add at least 10% of growth on the current GDP … For global trading… more than $30 trillion business… if we can add 10% more, it’s going to be $3 trillion add-on value to the whole system.” — Kuo Zhang (30:14)
- Inclusivity:
- Lower barriers will let more people and small businesses join global supply chains, compete, and grow
- Pursuing the mission to make business easy—for anyone, anywhere
Notable Quotes & Memorable Moments
- On lowering the barriers:
- “It’s completely lowering the barriers for people to enter into this field.” — Kuo Zhang (07:36)
- On scalability of agentic systems:
- “Now what we see is that they upload a file… hundreds or thousands of items, and you can just upload this file to Axio… It can be finished in hours or minutes.” — Kuo Zhang (08:49)
- On purpose-driven AI:
- “Whether your question is a real question or… a big enough question, I think that is the first one.” — Kuo Zhang (27:18)
- On impact of AI on global trade and GDP:
- “If we can add 10% more… it’s going to be $3 trillion add-on value to the whole system.” — Kuo Zhang (30:14)
Important Timestamps
- Intro and Alibaba.com background: 00:10–02:22
- Vision for global trade & technology’s role: 02:57–03:10
- Introduction to Axio: 04:49–06:05
- How Axio works (natural language behavior): 06:09–07:36
- Automation examples (Bolivian Games, ADHD apparel): 08:18–11:40
- Full lifecycle automation & agent capabilities: 11:40–13:23
- Handling qualitative vs. quantitative tasks, human-in-the-loop: 13:24–14:58
- Building AI for global compliance and nuance: 15:32–19:54
- User trust, guardrails, and transparency: 21:30
- Impact on SMEs & solo entrepreneurs: 22:14–25:09
- Technical and business challenges encountered: 25:22–26:57
- Advice for scaling AI products: 27:18–29:49
- Future economic impact and AI in global trade: 30:14
Resources for Further Exploration
- Visit: alibaba.com | xu.com
- Podcast: B2B Breakthrough by Alibaba.com — customer case studies, best practices (31:47)
- Events: “Co-Create” innovation summit (23:02)
Episode in a Nutshell
This conversation reveals the transformative power of agentic AI in global trade—making it faster, more accessible, and lowering the barrier to entrepreneurship at any scale. Axio, Alibaba.com’s AI agent, is built to handle the complex web of language, regulatory, cultural, and practical barriers that have historically challenged global B2B commerce. Kuo Zhang’s vision is clear: with the right mix of technology, platform, and culture, AI can democratize global trade, increase the world’s economic output, and fulfill the mission to “make business easy, anywhere, for anyone.”
