Lianlian DigiTech and UnionPay Partner on AI-Agent Cross-Border Payments

What was announced

On July 27, 2026, Hong Kong-listed Lianlian DigiTech (stock code 2598.HK) and UnionPay International — the international arm of China's state-owned card network — announced an expanded partnership to develop AI-agent payment capabilities for cross-border commerce. The deal extends an existing B2B payments cooperation into what the companies call "AI-agent payment scenarios."

The first concrete product: a human-in-the-loop AI-agent payment solution for global procurement. An AI agent will handle supplier matching, product selection, and payment execution within a single workflow — but the human buyer retains final approval and decision-making authority. The companies plan to refine the system based on operational feedback before expanding into additional sectors.

Lianlian DigiTech describes itself as an "AI-native company focused on building global financial infrastructure." Its AI agent, LoopXPay, already completed Greater China's first live B2B agentic transaction just three days earlier — on July 24 with Visa. In that test, the agent sourced a product sample, compared suppliers, placed the order, and executed payment — all within pre-defined spending controls and approval parameters. The transaction was supported by Visa's Agentic Directory, which registers verified AI agents for trusted interactions within Visa's ecosystem.

How it actually works

Let's be specific about what "AI-agent payments" means here, because the term is starting to get stretched. This is not a consumer chatbot that asks "shall I pay this invoice?" It is a structured procurement workflow where an AI agent operates within defined boundaries:

1. Supplier matching — the agent searches across suppliers, compares terms, and identifies candidates based on the buyer's criteria.
2. Product selection — it evaluates specifications, pricing, minimum order quantities, and lead times.
3. Payment execution — once the human approves, the agent executes the payment through UnionPay's global network, handling FX conversion, settlement routing, and compliance checks.

The explicit human-in-the-loop design is worth noting. Unlike some of the more speculative "fully autonomous AI commerce" pitches we have seen in the past year, Lianlian is positioning this as an assisted procurement tool — the AI does the legwork of finding, comparing, and paying, but the human signs off. That is both a practical choice (nobody wants an AI draining the company account) and likely a regulatory one.

The stated goals are concrete: reduce settlement timelines, minimize foreign exchange exposure, handle compliance requirements, and cut down manual reconciliation. For anyone who has dealt with cross-border B2B payments — where a single wire transfer can take 3–5 business days and cost EUR 15–50 in intermediary bank fees — the value proposition is immediately obvious.

UnionPay's global footprint — and what it means for Europe

UnionPay is the world's largest card payment processor by transaction value, overtaking both Visa and Mastercard in 2015. As of 2026, it claims acceptance in 183 countries and regions, with over 2.2 million merchants across 39 European countries — roughly 50% of all bankcard-accepting merchants in Europe, according to the company's own figures.

UnionPay cards are accepted by major European payment gateways including Adyen, BNP Paribas, and Stripe, meaning European online merchants can already process UnionPay transactions without additional infrastructure. The new partnership adds a B2B procurement layer on top of this existing consumer acceptance network.

For a Czech or German SME importing components from Shenzhen, this could shorten the payment chain significantly. Instead of routing through SWIFT with 2–3 intermediary banks, an AI agent could settle directly via UnionPay's network — potentially cutting settlement from days to hours and slicing FX spread costs.

A quick comparison of typical cross-border B2B payment routes to China:

The cost figures for the UnionPay agentic route are not yet published — Lianlian has not disclosed its fee structure — but the network economics suggest it will compete directly with fintech providers like Wise and Airwallex rather than traditional correspondent banking.

The regulatory elephant: AI Act and GDPR

Here is where the European angle gets interesting — and complicated. An AI agent that autonomously executes cross-border payments on behalf of a European business triggers at least two major EU regulatory frameworks.

The EU AI Act classifies AI systems used in financial services and payments as potentially high-risk, particularly when they involve automated decision-making with legal or financial consequences. While the human-in-the-loop design mitigates some of this risk, the Act requires conformity assessments, transparency documentation, and human oversight provisions — all of which Lianlian and UnionPay would need to address for European business users.

GDPR raises questions about data residency and processing. An AI agent handling procurement data — supplier identities, contract terms, payment amounts, bank details — for an EU-based company means personal and commercial data potentially flowing through Chinese infrastructure. Lianlian and UnionPay would need to demonstrate adequate safeguards under GDPR's cross-border transfer rules, likely via Standard Contractual Clauses or an adequacy decision (China does not currently hold one from the EU).

PSD3 / PSR — the EU's upcoming Payment Services Directive 3 and Payment Services Regulation — will introduce stronger requirements for payment service providers, including mandatory verification of payee (VoP) and stricter rules on open banking interfaces. An AI agent that initiates payments on behalf of a business would almost certainly fall within scope.

None of these issues are deal-breakers, but they mean European adoption will not happen at the same speed as in Greater China or Southeast Asia. Companies operating in Europe will need to build compliance layers — or wait for Lianlian and UnionPay to build them first.

Visa already got there first — this is a competitive move

The UnionPay announcement is better understood in context. Three days before the UnionPay deal was signed, Lianlian completed Greater China's first live agentic B2B transaction — on Visa's network, using Visa's Agentic Directory for trusted AI agent interactions. Visa's Global Head of Commercial Solutions, Darren Parslow, framed it as "helping advance the trusted foundations that businesses will need to participate in this next era of commerce."

Now Lianlian has brought UnionPay into the same game. The pattern is clear: Lianlian's LoopXPay AI agent is being positioned as network-agnostic middleware — it can route payments through Visa, UnionPay, and presumably other networks depending on cost, speed, and regional availability. This is the "financial infrastructure" play Lianlian's CEO Zhang Zhengyu described: identity verification, transaction authorization, intelligent payment routing, and global fund settlement as a unified service.

For the payment networks themselves, AI agents represent both an opportunity and a threat. On one hand, agent-driven payments mean higher transaction volumes and new use cases. On the other, if the AI agent is the one choosing which network to route through, the payment network's brand becomes invisible to the end user — it is just infrastructure.

Practical takeaway

For European businesses — particularly SMEs importing from or exporting to Asia — the practical implications are still mid-term rather than immediate. Lianlian and UnionPay have signed an agreement and announced a first deployment scenario; they have not launched a production service for European users yet.

What to watch:

• Pricing disclosure — when Lianlian publishes its fee structure for the UnionPay route, compare it against Wise, Airwallex, and traditional SWIFT. The FX spread will likely be the deciding factor.
• EU compliance roadmap — will Lianlian/UnionPay seek AI Act conformity assessment and GDPR certifications for European operations? Without these, EU companies may be unable to use the service for anything beyond non-sensitive procurement.
• Competitive response — Mastercard and SWIFT are not sitting idle. Mastercard recently expanded its virtual card platform with embedded ecosystem partnerships, and SWIFT has been piloting AI-assisted compliance screening. Expect announcements from the Western networks within months.

As for the fully autonomous AI procurement agent that handles everything from sourcing to payment without human intervention — that is still firmly in the demo phase. The human in the loop is not going anywhere, and for good reason.