The first live B2B agentic transaction in Greater China shows how AI agents are beginning to move from administrative tools into operational roles. In the transaction, LianLian’s LoopXPay agent—with support from Visa’s agentic commerce solutions—managed a supplier payment process within defined spending and approval parameters.
The agent autonomously mapped the purchasing requirement, suggested potential suppliers, compared options, placed the order, and executed the payment within a unified workflow.
The objective of the pilot was to show how AI agents could reduce the complexity of commercial payments, especially for small- and medium-sized businesses. By taking on time-consuming operational tasks, AI agents could allow business owners to spend more time on growth and strategic priorities.
Automating Pain Points
Streamlining commercial payments flows has been a priority for many organizations, but it is especially important for small businesses that often operate with limited resources.
Accounts receivable and accounts payable (AR/AP) processes have long been a pain point for these organizations because they require constant human involvement in such tasks such as managing credit lines, reviewing invoices, reconciling payments, and resolving exceptions.
Artificial intelligence offers substantial potential to streamline and automate many of these functions. This is why businesses have made significant investments in generative AI, and agentic AI has the potential to accelerate automation even further.
An Emerging Model
Moving agentic transactions from pilot programs into real-world business environment will require organizations to address several operational and security considerations.
Implementing agentic AI securely and responsibly within critical workflows will require infrastructure with strong governance and oversight controls, while also enabling orgnizations to deploy and refine these systems iteratively.
Given the complexity of many AR/AP processes, these systems have also become frequent targets for fraud, partly due to AI-driven schemes developed by cybercriminals. While many of these attacks are designed to manipulate humans, criminals will likely adapt their tactics to target AI agents deployed within organizational systems.
Beyond security concerns, there are also remaining gaps in the agentic commerce model that the industry is still working to address. One of the most critical challenges is that AI agents lack a native, secure way to transact and interact with one another in a system originally designed for human users.
Despite these challenges, the potential of AI in all its forms is too significant for businesses to ignore. As more trials are conducted worldwide, the agentic commerce ecosystem will continue to evolve, helping define how AI-driven transactions operate at scale.
