Agentic AI has the potential to transform payment systems and e-commerce by shifting from basic automation to autonomous systems capable of complex reasoning and task management.

While the technology promises to enhance user experience and personalization, improve operational efficiency, and support cost reduction, it also introduces risks and challenges, including threats to market stability, data security vulnerabilities, and regulatory complexities surrounding non-human authentication and accountability, according to a note by the International Monetary Fund (IMF).

Agentic AI capabilities in payments

AI agents are set to change e-commerce by replacing simple automation with autonomous reasoning. Unlike basic scripts that follow fixed rules, these agents can think through problems and manage complex tasks on their own.

In online commerce, shoppers will benefit from personalized decision support that incorporates preferences, constraints, and real-time price signals. This personalization can increase customer lifetime value through more relevant product matching and streamlined purchasing experiences.

Agentic systems may also extend beyond checkout to post-purchase functions, including delivery coordination and returns management. Gartner projects that by 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, leading to a 30% reduction in operational costs.

For merchants, agentic AI will modernize operations. These systems enable seamless integration between different AI tools, allowing for smoother operations, including automated return processing and dynamic pricing adjustments that help businesses lower service costs, resolve disputes more effectively, and react instantly to changes in inventory or demand.

Beyond smart automation, AI agents can coordinate and execute multistep workflows across financial distributed networks autonomously. In cross-border payment, an AI agent can orchestrate the entire payment payment chain from payment initiation, optimizing routing options, triggering compliance checks, and monitoring settlement and post settlement exceptions. Proponents believe that this automated payment flow can reduce delays associated with manual intervention and rigid workflows.

Additionally, AI agents can help streamline foreign exchange (FX) management, allowing financial institutions and companies to continuously monitor real-time exchange rates, analyze spreads across banking rails, optimize timing for conversion, and choose cost-effective paths for transferring funds across multiple currencies. Through predictive analytics, they can also support improved forecasting and risk management functions in areas such as liquidity planning and FX management, generating significant cost savings and operational efficiencies.

Finally, AI agents have the potential to significantly improve compliance processes by embedding regulatory logic directly into operational workflows. Unlike traditional automation tools, agentic systems can interpret objectives, monitor activity in real time, and autonomously take actions within predefined guardrails such as flagging suspicious transactions, escalating high-risk cases, or adjusting controls when regulatory thresholds are met.

This allows compliance functions to operate at the same speed and scale as modern digital systems, reducing operational burden and human error while strengthening consistency, traceability, and regulatory alignment.

Risks and challenges associated with agentic AI

While agentic AI promises numerous benefits and efficiency gains, the autonomy, opacity, and non-deterministic behavior of these systems introduce material risks to consumer protection, market stability, and regulatory oversight.

A primary concern is that agentic systems may misinterpret user intent, optimize provider incentives rather than user welfare, drift from original objectives over time, or engage in subtle behavioral nudging at scale.

Another concern is algorithmic herding. If dominant models identify identical market signals, they may act simultaneously, bypassing traditional safety mechanisms like circuit breakers and triggering flash crashes. In payments, this can impair the functioning of payment systems by synchronizing liquidity demand, amplifying procyclical behavior, and creating congestion across payment rails, thereby undermining the predictability and resilience of core financial market infrastructures.

Additionally, generative AI (genAI) models are prone to “hallucinations,” creating false information with high confidence. In financial contexts, such errors can have significant consequences.

Another significant risk relates to data security and privacy protection. Autonomous agents relying on third-party services such as cloud providers, AI model endpoints, and financial services require user’s sensitive data such as bank credentials, credit card numbers, and crypto wallet keys. This exposes the user to data leaks and privacy concerns and creates a highly concentrated point of vulnerability.

There are also concerns about market concentration and competition. GenAI is fed by vast amounts of data which require computing power that can only be provided by a few, dominant companies. This has led to a highly concentrated AI supply chain, ranging from data centers to cloud computing and AI applications. Such concentration can threaten innovation and raise financial stability, operational, and reputational risks.

Besides direct risks, there are gaps that hinder the adoption of agentic AI in payments. A major issue is the growing skills gap in the financial sector, with a shortage of professionals who possess both financial expertise and advanced AI knowledge.

A 2025 study by OneStream emphasizes this issue. The survey, which polled more than 2,500 corporate finance professionals, found that 57% of current professionals believe a generational technology divide is a pain point within their organizations. Of those acknowledging a tech divide exists, the AI skills gap (44%), the pace of technology changes (44%), and attitudes towards AI replacing tasks (40%) were named as the top contributes to this gap.

Financial institutions also face technical hurdles in integrating these new technologies into their existing environments. Many banks still relay on legacy infrastructure, which may not be compatible with cutting edge AI technologies. This necessitates substantial investments in system upgrades and data migration.

Finally, specific gaps remain regarding authentication or know-your-customer (KYC) requirements for agents. Traditional authorization mechanisms, including KYC processes and multifactor authentication, are designed around human users who explicitly approve transactions. When payments are initiated autonomously by software agents acting under delegated authority, verifying both the identity of the agent and the intent of the underlying user becomes significantly more complex. This raises questions not only around authentication, but also around accountability and compliance.

The state of agentic AI adoption

Though adoption of AI agents in payments remains at an early stage, significant experimentation is underway. Additionally, major technology firms, payment networks, and financial institutions are actively developing the fundamental technologies and protocols needed to integrate these autonomous systems into daily commerce.

OpenAI launched in September 2025 Instant Checkout, powered by the Agentic Commerce Protocol. The feature allows users to buy eligible products directly inside ChatGPT by tapping “Buy,” confirming shipping and payment details, and completing the purchase without leaving the chat.

Google launched in January 2026 the Universal Commerce Protocol (UCP), a new open standard for agentic commerce that works across the entire shopping journey. UCP establishes a common language for agents and systems to operate together across consumer surfaces, businesses and payment providers.

Amazon introduced in March 2026 Alexa for Shopping, a personalized, agentic AI assistant on Amazon. The solution combines deep product knowledge, in-depth information from across the web, and powerful shopping capabilities with a user’s personal preferences, shopping history, and conversations from across both Amazon.com and Alexa.

Across the broader financial industry, the deployment of AI agents is expected to grow rapidly in the coming years. A 2026 report led by the University of Cambridge surveyed in late 2025 over 600 firms and regulators worldwide, and found that agentic AI is already in active adoption among 52% of industry respondents. Looking ahead, 81% believe that agentic AI will be meaningfully achieved by 2030. This suggests that agentic AI represents the most promising growth frontier in current AI technology.

Featured image: Edited by Fintech News Switzerland, based on image by freelancerdesign098 via Magnific