Why AI agents will be a global trade game changer for SMEs

The agents are coming. 2024 was the year when artificial intelligence (AI) and large language models (LLMs) became ubiquitous and the subsequent competition among numerous LLMs has both substantially reduced the cost of models quickly and narrowed the performance gap between them, as illustrated by the latest DeepSeek’s challenge of OpenAI’s supremacy.

However, such generalized AI – or “horizontal AI” – tends to lack the analysis and customization needed for complex business challenges.

Now in 2025, “vertical” AI agents – purpose-built AI tools designed for specific industries – represent the next logical step in AI technology development. Microsoft, Google and Nvidia have all begun positioning and pushing so-called "agentic" offerings.

Unlike general-purpose AI systems like ChatGPT, agentic AI refers to self-governing software programmes that perceive their industry environment, make business decisions and act to achieve specific goals.

Take the global supply chain, for example. What are the main challenges for small and medium enterprises (SMEs), especially those in emerging markets?

In the digital era, B2B global trade has been defined by the current digital age where SMEs can access the unlimited network of the global supply chain online, make digital payments, utilize translation technology and track their orders. Despite these advancements, SMEs face significant information overload throughout the sourcing process, for which AI agents could excel.

According to United Nations Trade & Development’s projection in December, global trade is on track to hit a record $33 trillion in 2024. The global trade market is vast, and SMEs need to spend increasingly more time to verify, compare and make decisions. In addition, some of the steps to complete a transaction are especially challenging:

The traditional solution to those challenges is hiring experts and agencies. Yet while hiring someone with expertise sounds simple, the downside is that a person’s connections and resources are also limited – plus, not everyone can afford them.

As such, global trade – complex, dynamic, repetitive – is the ideal setting for agentic AI. The expertise in industries like supply chains – i.e. the “vertical” depth of related AI agents – highlights the substantial distinction between AI agents and AI assistants.

AI assistants are task-based, but AI agents are goal-based. AI assistants focus on completing specific tasks given direct instructions, while AI agents are designed to autonomously achieve broader goals.

The new AI agents are digital colleagues. They help you plan, problem-solve and act to achieve a goal so that everyone can become a pro. By simplifying sourcing processes and saving time and resources, AI can create impactful change for SMEs who are already in, or just entering, the global sourcing landscape.

For example, e-commerce platform Alibaba International launched its AI agent, Accio, as an AI-powered B2B (business-to-business) search engine for product sourcing. According to Alibaba’s data, during the peak global e-commerce sales seasons in November and December 2024, more than 50,000 SMEs worldwide actively used Accio to source inspirations for Black Friday and Christmas inventory stocking.

Different from traditional keyword search engines, Accio uses natural language processing to enhance search capabilities, enabling SME buyers to intuitively find the products they need. It now boasts more than 500,000 SME users and, impressively, half of the top 20 countries using this tool are SMEs in developing countries, such as Pakistan, Vietnam and Bangladesh.

If you have a clear idea of what you are looking for, Accio can quickly summarize industry knowledge and actively guide you step by step. But what’s more useful is when you have an open-ended question, for example, "looking for business ideas for Christmas sales campaign". The AI agent gathers information from across the internet, offers you business insights and helps you to dive deeper in a series of open-ended questions, until it reaches a concrete plan based on supply information of key ingredients.

In the Harry Potter series, Accio is the summoning charm that summons objects towards the caster over a long distance. Similarly, the AI tool can then synthesize the information into a request for quotation (RFQ) that can then be issued to potential sourcing partners, simplifying the typically complex and time-consuming RFQ process for business owners.

What about "hallucination", when a LLM generates nonsensical or inaccurate outputs? After all, just think of all the ways a hallucination could screw up buying airplane tickets without careful oversight. That’s where the specific industry knowledge becomes critical. In the case of Accio, the 25-year global e-commerce data accumulation of Alibaba.com and more than one billion existing product listings are the distinguishing factor.

To ensure that all information provided by the AI sourcing agent is useful, Alibaba.com has tapped into retrieval-augmented generation features. These features filter search results to information only from sources deemed reliable by the AI tool according to a knowledge base. According to Alibaba.com, if the AI tool is unsure about something, it is programmed to acknowledge that, rather than provide unsubstantiated guesses.

Now, as illustrated by the example of Accio, which still needs to evolve, we could see three critical elements of vertical agentic AI – specialized AI systems designed to streamline enterprise functions – to differentiate from mere chatbots:

In summary, while the first iteration of AI copilots augmented human tasks, the next generation agents could fundamentally change how businesses operate.

For SMEs, AI represents a powerful tool for data-driven insights, supply chain optimization and personalized customer engagement, enabling them to compete globally.

Accordingly, tools like sourcing agents have to the potential to revolutionize the cross-border trade game around the world, for businesses of all sizes to thrive.