The Convergence of Open Finance and AI -Part II: The Rise of Agentic AI: Beyond Assistance to Autonomy

1. Introduction

While AI applications in data analysis and automation have already transformed financial services, the industry is now entering a new phase with the emergence of agentic AI. This represents a significant evolution from systems that merely analyze data or generate content to those that can autonomously act, react, learn from feedback, and pursue goals with minimal human intervention.

Agentic AI systems possess "agency," the capacity to make decisions, take actions, and interact with external environments independently. Key characteristics distinguishing these systems include autonomy (performing tasks without direct human oversight), adaptability (learning from interactions and feedback), goal orientation (reasoning about steps to achieve objectives), perception and reasoning (gathering and analyzing environmental data), and action execution (interacting with external systems to complete tasks).

2. Defining Agentic AI

Key characteristics distinguishing agentic AI include:

• Autonomy: Performing tasks without direct human oversight or step-by-step direction.16

• Adaptability: Learning from interactions, environmental feedback, and past experiences to modify future decisions and actions.

• Goal Orientation: Receiving specific tasks or objectives and reasoning about the steps required to achieve them.

• Perception and Reasoning: Gathering data from their environment (databases, sensors, APIs), identifying patterns, and using reasoning engines (often powered by Large Language Models - LLMs) to understand context and formulate plans.

• Action Execution: Interacting with external systems via APIs or other interfaces to execute planned tasks, often within predefined guardrails.

These systems effectively combine the flexibility and natural language understanding of LLMs with the precision and reliability of traditional programming, allowing them to handle complex, multi-step workflows that were previously challenging for AI.

3. Agentic Commerce: The Next Frontier

The application of agentic AI to commerce represents a fundamental shift in how consumers interact with financial services. Instead of actively browsing, comparing, and purchasing, consumers can delegate these tasks to autonomous AI agents.

These agents are designed to predict needs by analyzing user preferences and contextual data, discover products by intelligently searching across platforms, negotiate terms by tracking price fluctuations and identifying discounts, execute purchases by completing entire transaction processes, and manage post-purchase tasks like shipment tracking and returns.

This evolution signifies a move beyond AI assisting commerce to AI conducting commerce, representing what analysts term a "seismic shift" comparable to the advent of eCommerce and mobile commerce.

"Commerce, everything we buy and sell, represents a whopping $2.7 trillion part of our economy – and it’s almost always growing. Part of what fuels that growth is technology investments made by retailers and brands in modern capabilities that allow them to reach consumers in more flexible, immersive and personalized ways. Over $200 billion is invested annually on these types of technology innovations – and by the early 2030s, Gartner predicts spending on commerce tech will reach over $400 billion." Scott Friend - Bain Capital Ventures[1]

3.1 The Payment Giants Embrace Agentic AI

Recognizing the transformative potential of agentic commerce, the world's largest payment networks and processors, Visa, Mastercard, and PayPal, have moved swiftly to position themselves at the center of this emerging ecosystem[2].

Visa's "Intelligent Commerce" (VIC) initiative aims to establish a new standard for the AI commerce era by opening its global network (VisaNet) to developers building AI agents. The core technological component is the "AI-Ready Card," a tokenized digital credential linked to consumers' underlying Visa accounts that can be provisioned to their chosen AI agents.

Mastercard's "Agent Pay" solution is part of a broader Agentic Payments Program designed to integrate AI and payments securely. Similar to Visa, Mastercard's approach centers on "Agentic Tokens," which build upon existing tokenization capabilities already used in digital wallets, secure card-on-file solutions, and programmable payments.

PayPal's approach differs from the card networks, focusing on enabling developers to build agentic experiences directly on its platform through an "Agent Toolkit" and access tokens that allow AI agents to interact with the PayPal platform via APIs.

3.2    Agentic AI and Open Banking: Competition and Synergy

The relationship between agentic commerce payment initiatives and Open Banking presents both competitive tensions and opportunities for synergy.

3.3 Areas of Competition

The most direct area of conflict is in payment initiation. Agentic payment solutions from Visa, Mastercard, and PayPal provide alternative ways for AI agents to execute payments. If these methods, particularly tokenized cards integrated seamlessly into digital wallets, become the default for agentic transactions due to ease of setup or existing user trust, they could significantly hinder the adoption of Open Banking A2A/Pay-by-Bank for this growing segment of e-commerce.

This scenario would reinforce the dominance of existing card networks and large payment platforms, counteracting one of Open Banking's goals of fostering competition in payments. The inherent friction sometimes associated with Open Banking PIS (Payment Initiation Services) flows (such as multiple authentication steps) could make tokenized or platform-API payments appear more seamless for autonomous agents, despite potential higher costs for merchants.

3.4 Areas of Synergy

Agentic AI thrives on data to deliver its core value proposition of hyper-personalization and intelligent decision-making. Open Banking Account Information Services (AIS) provides a standardized, regulated, and consumer-permissioned channel to access rich, cross-institutional financial data. AI agents designed for financial management, budgeting, or providing personalized shopping recommendations could significantly enhance their capabilities by integrating Open Banking AIS data feeds.

Agentic AI could potentially improve Open Banking processes themselves, for instance, by enhancing security through intelligent fraud detection layered on top of Open Banking transactions, automating consent management, or providing smarter analysis of aggregated data.

4.  Reshaping Consumer Money Management

The integration of agentic AI with rich data sources, potentially including Open Banking, is fundamentally reshaping how consumers manage their finances. The core promises are hyper-personalization and convenience, moving beyond traditional tools towards truly autonomous financial assistance.

Agentic AI systems excel at analyzing vast amounts of data, transaction histories, spending patterns, stated goals, possibly even contextual information, to understand individual financial situations and preferences with unprecedented granularity.

This allows for tailored recommendations (moving beyond generic advice to highly specific suggestions based on individual spending patterns), proactive insights (pushing relevant information rather than requiring manual analysis), and automated actions (functioning as digital concierges for finding deals, making purchases, and managing financial tasks).

3.1 The Evolution of Personal Finance Management (PFM) Tools

Agentic AI marks a significant evolution from traditional PFM tools. Historically, PFM applications primarily focused on aggregating financial data and providing retrospective reporting and budgeting capabilities, often requiring considerable manual input from users. They operated on a "pull" model, where users needed to actively seek out information.

Agentic AI transforms PFM into a proactive, "push" model. These new tools aim to automate money management, provide actionable advice, simplify complexity, and continuously learn from user behavior. This evolution represents a shift from tools that help users understand their finances to agents that help manage and optimize them, often autonomously.

4.  Critical Challenges: Trust, Security, and Regulation

The transformative potential of agentic commerce is tempered by significant challenges related to security, data privacy, consumer control, liability, and the evolving regulatory landscape.

Introducing autonomous AI agents into the payment ecosystem creates new security considerations. While tokenization reduces risks associated with data breaches, vulnerabilities could exist in token issuance, storage, or validation processes. For API-based approaches, risks include insecure API keys/tokens, inadequate authentication or authorization logic, and the vulnerability of third-party providers accessing the APIs.

"Recent advances in AI agents could expand the utility of these technologies across the private and public sectors, but they also raise many data protection considerations. While practitioners may be aware of some of these considerations due to the relationship between LLMs and the latest AI agents, the unique design elements and characteristics of these agents may exacerbate or raise new compliance challenges." Daniel Berrick - Senior Policy Counsel for Artificial Intelligence, Future of Privacy Forum[3]

A major area of uncertainty is liability for unauthorized or erroneous transactions initiated by an AI agent. Existing consumer protection laws were not designed with autonomous AI agents in mind. Key questions include whether a transaction is "unauthorized" if the user granted the agent broad permissions but it acted erroneously, how the legal doctrine of agency applies when the agent is an AI, and who bears the financial risk between consumers, AI providers, merchants, banks, and payment networks.

Currently, there is no specific, comprehensive regulatory framework governing agentic AI, leading to uncertainty for developers, providers, and users. Relevant existing regulations include electronic transaction laws, data privacy frameworks, payment services regulations, and emerging AI-specific regulations like the EU AI Act. There is a clear need for collaboration between industry stakeholders and regulators to develop clear standards and guidelines for agentic commerce.

5. Strategic Recommendations for Financial Institutions

Navigating this transformative landscape requires strategic adaptation from all stakeholders in the financial ecosystem:

5.1 For Financial Institutions:

• Develop a dual strategy addressing both internal applications of agentic AI (automating compliance, risk management, operations) and participation in the external agentic commerce ecosystem.

• Decide whether to primarily enable agentic payments via existing rails (supporting tokenization requests from networks) or to actively compete by enhancing Open Banking APIs to be more agent-friendly.

• Utilize Open Banking AIS data (with consent) to power proprietary AI-driven personalization, PFM tools, and advisory services.

• Prioritize building robust security, privacy, and governance frameworks for AI interactions to maintain customer trust.

"The rise of agentic traffic presents unique challenges—and opportunities—for financial institutions and fintechs. How can you make your agent 'discoverable' among millions of others? How will you ensure your agent is 'up-to-date' with your latest product offerings? Moreover, managing the technical infrastructure to support both traditional developers and emerging AI agents could quickly become overwhelming. To stay ahead, financial organizations must invest now in establishing a robust single source of truth—effectively managing their tech stack and driving adoption across the spectrum, from developer-focused SDKs to up-to-date MCP servers optimized for AI agents." Adeel Ali - CEO, APIMatic

5.2 For Fintechs and Third-Party Providers:

• Focus on building specialized AI agents for specific financial tasks (advanced PFM, lending analysis, compliance) or providing critical enabling infrastructure.

• Leverage Open Banking AIS data to fuel sophisticated agentic PFM and advisory tools.

• Strategically position offerings relative to large payment players, collaborating where necessary but differentiating through innovation and user experience.

• Prioritize creating seamless, intuitive, and trustworthy agentic experiences.

5.3 For Merchants:

• Prepare digital storefronts and backend systems for interaction with AI agents, not just humans.

• Adapt marketing and customer acquisition strategies for a world where AI agents act as intermediaries.

• Assess the costs, benefits, and integration requirements of supporting various agentic payment methods.

6.  Conclusion: Embracing the Agentic Future

The convergence of Open Finance and AI technologies, particularly agentic AI, represents a paradigm shift in how financial services are delivered and experienced. The initiatives by Visa, Mastercard, and PayPal to enable agentic commerce mark a pivotal moment in this evolution, signaling the industry's recognition of autonomous AI's transformative potential.

For financial institutions, the path forward involves strategic decisions about how to participate in this new ecosystem, whether by embracing tokenization, enhancing Open Banking APIs, or developing proprietary AI agents. The organizations that successfully navigate the tensions between existing payment rails and Open Banking frameworks, while addressing the critical challenges of security, privacy, and liability, will be best positioned to thrive.

The future of finance isn't just open; it's intelligent and increasingly autonomous. By understanding the complementary strengths of Open Finance data access and AI capabilities, forward-thinking institutions can build the next generation of financial services that are more personalized, efficient, and accessible than ever before.

As we enter this agentic era, one thing is clear: the institutions that view Open Finance and AI not as separate initiatives but as complementary forces will gain significant competitive advantages in the rapidly evolving financial landscape.

[1] https://baincapitalventures.com/insight/the-dawn-of-the-agentic-commerce-era/

[2] https://www.retailtouchpoints.com/topics/payments/visa-mastercard-paypal-dive-into-agentic-era-with-tools-that-help-consumers-use-ai-to-buy

[3] https://fpf.org/blog/minding-mindful-machines-ai-agents-and-data-protection-considerations/