Who Will Dominate the Next Generation of Smart Agent Access Points and Ecosystems?

/

By Zheng Jiuyu | Edited by Yang Xiaoruo

Produced by Business Show

The AI payment battle was set ablaze by two tech titans within a mere 48 hours.

On June 17, WeChat Pay officially unveiled an AI-specific card integrated within WeChat Balance, tailored for AI smart agent payment scenarios and linked to Tencent's desktop productivity smart agent, WorkBuddy.

Users can navigate to the WorkBuddy expert center page, access services like the Meituan Life Assistant, and inquire about deals such as 'What are some good group-buying deals nearby?' After the smart agent recommends products, users can bind the AI-specific card, top it up, and complete the purchase by confirming the deduction from the exclusive card via their mobile device. This marks a significant leap in AI's transition from efficiency tools to consumer-facing applications.

Just a day prior, Alipay rolled out an invitation-only trial of its AI iteration, 'Abao.' Compared to its predecessor, the AI-driven Alipay doesn't boast groundbreaking features, presenting a straightforward interface with merely two functional pages: Abao and Assets. Users can engage with the smart agent to access functionalities like asset allocation advice and life services. However, for financial transactions such as self-directed trading and wealth management, Abao cannot execute these operations directly. Nevertheless, it can manage daily life services such as ride-hailing, ticket purchases, and food delivery.

Ultimately, all operations redirect to Alipay mini-programs, mirroring Alibaba's AI assistant, Qianwen, albeit with varying underlying ecological coverage capabilities.

As payments evolve from scanning codes to conversational interactions and the decision-making chain extends from humans to AI smart agents, the battle for the primary access point into financial and other consumer scenarios in the AI era unfolds discreetly behind the scenes of 'voice-activated payments.'

01 The 'Under Construction' Phase of AI Payments

After exploring Alipay's AI iteration 'Abao' and WorkBuddy's AI-specific card, it's evident that these AI payment products are still in the 'under construction' phase.

Firstly, challenges persist at a macro level regarding cross-ecosystem coordination.

Despite WeChat and Alipay boasting massive user bases—Tencent's Q1 2026 financial report, released in May 2026, revealed that WeChat and WeChat Pay combined had 1.432 billion monthly active accounts, while public data indicates Alipay has over 1 billion users—their respective mini-program ecosystems and merchant systems are not fully interoperable. When users attempt to access third-party services like Meituan within smart agents such as WorkBuddy, they often encounter payment gateway hurdles.

Due to incomplete adaptation of third-party merchants' payment interface configurations with WeChat's AI-specific card, users ultimately need to redirect to the original app or mini-program to complete payment via scanning. In complex cross-platform ecosystems, current AI payment channels remain fragmented, with significant gaps between merchant-side payment configurations and AI smart agents.

Secondly, AI large models still grapple with severe inaccuracies in financial and consumer scenarios. Payment and financial analysis scenarios demand impeccable data accuracy, but current large language models (LLMs), essentially probabilistic prediction models, struggle to entirely eliminate inaccuracies.

Abao's Assets section integrates various funds within Alipay, categorized into three types: liquid assets, wealth management assets, and credit assets, corresponding to Yu'ebao and bank cards (savings and money market funds), gold investment plans (wealth management products), and credit cards and Huabei (credit products), respectively.

However, when analyzing bills and assets involving users' vital interests, AI may exhibit data omissions, calculation errors, or even misclassify consumption from non-associated platforms.

While AI's fabricated responses may be inconsequential in casual Q&A, in the financial payment sector, such algorithmic black-box-induced data distortions undoubtedly erode user trust in the platform.

Additionally, AI's rigidity in understanding user intent remains apparent. When users express vague consumption needs, current AI often mechanically invokes the search interface of food delivery mini-programs without comprehending the user's true intent for personalized recommendations, necessitating users to articulate specific demands.

This suggests that AI in current payment scenarios functions more like a precise triage desk, far from the role of a true personal assistant.

However, comparing AI payments with traditional payment chains unveils revolutionary disparities. Traditional payments necessitate users to actively seek services—opening apps, searching keywords, browsing and comparing prices, adding items to the cart, clicking checkout, and finally entering passwords to confirm. At each step, users may hesitate or abandon the process.

In contrast, AI payments reconstruct these steps as services actively seeking users. For instance, within WeChat's WorkBuddy ecosystem, a casual query like 'What's good to eat nearby?' enables AI to instantly search surrounding restaurants, compare reviews, generate recommended orders, with users only needing to confirm.

Similarly, in Alipay and Qianwen's ecosystems, whether ordering a coffee or booking a flight, AI can dissect complex consumption decisions and execute them, enhancing decision-making efficiency.

However, AI payments are not merely about adding a voice-activated interface to Alipay or WeChat. Essentially, they embed payment capabilities into the service chain behind the entire smart agent ecosystem, though some products' voice recognition experiences remain less seamless and require further refinement.

In this AI payment revolution where conversations equate to transactions, the ultimate test lies in who can proactively and accurately understand and grasp user intent while seamlessly executing the underlying service processes.

02 Who Will Dominate the Next Generation of Smart Agent Access Points and Ecosystems?

From a timing perspective, some may perceive WeChat Pay's launch of the AI-specific card as a defensive move against Alipay's AI iteration.

In reality, on April 9, WeChat Pay already introduced an AI payment access capability system, including Skill packages, AI-friendly documentation, and APIs, enabling developers to generate payment codes using natural language descriptions. Simultaneously, it disclosed that over 70% of merchant developers utilized AI-assisted programming tools for access development.

Not just WeChat Pay, but on February 11 this year, JD Technology also unveiled a novel payment method, 'JD AI Pay,' powered by its self-developed JoyAI large model, allowing users to complete payments via voice commands. This feature was initially implemented in terminals like the JoyAI App and smart glasses JoyGlance, achieving a closed loop of 'glance to order, voice to pay.'

Thus, in the AI payment race, companies are competing for the conversational transaction payment traffic access point, attempting to shift users from 'active search and order' to 'one-sentence intent-driven' AI smart agent commercialization.

An industry insider informed Business Show that WeChat's true concern may stem from the rise of independent AI applications like Qianwen and other general-purpose large models and smart agents. If users become accustomed to 'turning to AI first' for tasks like 'ticket search-price comparison-food delivery-payment,' WeChat would be relegated to a mere underlying payment channel, losing the primary access point for direct user interaction.

Since January this year, the Qianwen App has fully integrated with Alibaba's ecosystem, including Taobao, Fliggy, Gaode, and Alipay, launching over 400 AI service functions. By connecting with Alipay's native 'AI Pay,' Qianwen achieved one-click order and payment within conversations, entering the one-sentence service phase.

From this vantage point, WeChat Pay's launch of the AI-specific card may reflect a deeper consideration to firmly secure payment and account defenses within the smart agent ecosystem.

However, while vying for this access point, WeChat continues to exhibit extreme caution. WeChat's AI-specific card is essentially an isolated fund pool. According to official announcements, it is completely isolated from WeChat's main account, with AI only able to access the card's balance, and each transaction requiring user password or fingerprint confirmation. This dedicated fund, transaction-by-transaction confirmation design reflects WeChat's prudence regarding AI autonomy, positioning AI as a service wallet that prioritizes addressing trust cost issues.

After all, AI payments involve financial industry operations, necessitating primary consideration of regulation and compliance. To prevent AI from overstepping authority and misusing funds, some degree of convenience must be sacrificed to uphold fund security.

Compared to WeChat, Alipay's Abao pursues a full-stack ecological aggressive route, deeply embedding AI within the app to cover life, government affairs, and even financial asset management, attempting to establish a seamless closed-loop commercial ecosystem.

Whether WeChat's caution or Alipay's aggressiveness, different approaches fundamentally represent explorations of the 'AI autonomy' boundaries. Industry observers believe they are vying for industry standards and dominance.

Epilogue

It is evident that the full-scale eruption of AI payments hinges on trust mechanisms and responsibility boundaries.

Although the industry pursues frictionless payments, most transactions still require password verification or manual confirmation based on current experiences. This represents the industry's current challenge: decision unexplainability due to algorithmic black boxes and ambiguous responsibility attribution.

If AI misinterprets intent or overcharges during multi-step operations, who bears the responsibility—the user, AI platform, or merchant? This poses significant security risks.

Meanwhile, before trust mechanisms are perfected, frictionless payments face substantial psychological barriers, indicating the industry must undergo a user education phase.

AI payments are not solely about payment; with AI capabilities, whether products and platforms succeed ultimately depends on establishing truly trustworthy transaction execution and service capabilities. Security, compliance, and risk control remain the Sword of Damocles hanging over all AI payment products. "The End"