Can AI Autonomous Agents Disrupt Digital Business Models?

Dominic Jainy stands at the forefront of the most significant shift in digital commerce since the invention of the smartphone. As an IT professional with deep roots in artificial intelligence and blockchain, he has watched the transition from simple automation to full autonomous “agency” with a critical eye. Currently, we find ourselves in a high-stakes standoff between consumers who crave the frictionless convenience of AI agents and enterprises that are terrified of the security and liability risks these agents introduce. Dominic’s perspective is shaped by years of integrating complex machine learning models into enterprise workflows, making him a vital voice for organizations trying to navigate this “agentic” revolution without falling into a legal or technical abyss.

The following conversation explores the fundamental friction between current web architecture and the autonomy of modern software. We delve into the “circular firing squad” of liability that emerges when AI makes a financial error, the reason why major banks are currently aggressive in their blocking of third-party tools, and the specific four-step roadmap enterprises must follow to adapt. The dialogue moves past the technical jargon to address the human and legal realities of 2026, where the “user” at the keyboard is increasingly a line of code rather than a person.

Many financial institutions are currently detecting and blocking AI agents that use customer-provided credentials, creating a significant point of tension. Why is this confrontation happening now, and why are banks so hesitant to embrace this kind of automation?

The friction we are seeing stems from a total misalignment of expectations between the consumer and the institution. On one side, you have users who feel they are being incredibly slick by using AI agents to chase better interest rates, pay bills automatically, or optimize their subscriptions across multiple platforms. To the customer, this feels like the ultimate realization of the cloud’s promise—an intelligent layer that removes the mental load of financial management. However, when these agents log in using customer credentials, they are often violating the bank’s strict terms of service which prohibit non-human third parties from operating accounts through consumer-facing websites. Banks are looking at their telemetry and seeing activity that they cannot definitively identify as human, and that uncertainty triggers every red flag in their security protocols. They aren’t just being stubborn; they are struggling to tell the difference between a helpful AI agent and a sophisticated credential-stuffing attack or a security incident. This has led several major US banks to take a hardline stance, as reported by Axios, where they are actively restricting or outright blocking these outside agents to protect the integrity of their platforms.

Most of our digital infrastructure was built under the assumption that a human would be the primary operator. How is that fundamental design of the web clashing with the rise of autonomous software today?

The web was essentially designed around the “human at the keyboard” abstraction, assuming that the person owning the account is the one clicking the buttons. We’ve seen this assumption slowly erode over the years with screen scraping and basic browser extensions, but AI agents have completely shattered it. Unlike a simple budgeting app that just reads your balance, a modern AI agent has true “agency”—it can reason across tasks, handle ambiguity, and actually trigger contractual obligations or move funds. This is a massive leap from a brittle script or a scheduled batch job; an AI can see a balance, compare it against a high-yield account elsewhere, and decide to move the money while simultaneously opening a support chat to question a fee. The interfaces we use, which were designed for sensory human interaction, are being treated as data sets and command lines by autonomous software. This mismatch creates a scenario where the “agent” is impersonating a human to get work done, which is a fundamentally unstable foundation for digital business.

When an AI agent makes a mistake—perhaps moving money incorrectly or canceling the wrong service—who is ultimately responsible in this “circular firing squad” of liability?

The liability issue is perhaps the most daunting “nightmare scenario” facing enterprises today because there is no clear governance model for autonomous mistakes. If a transaction goes sideways, the customer will naturally claim they never authorized that specific, incorrect action, while the bank will point to their policies stating that handing credentials to a third party is a violation. Meanwhile, the agent platform will argue they followed the user’s broad configuration, and the cloud provider will distance themselves by saying they only provided the raw infrastructure. In many regions, consumer protection rules are heavily weighted in favor of the customer, meaning a bank might be legally forced to unwind a disastrous transaction even if it was triggered by an unsanctioned agent. This creates a terrifying imbalance where the enterprise bears the financial downside of automation without having the control plane to manage the risks involved. It isn’t just a banking problem either; whether you are an airline, an insurer, or a healthcare provider, if your site allows for meaningful action, you are in the line of fire.

Given the risks, what immediate steps should enterprises take to ensure they aren’t just reacting to these agents but are actually building a sustainable strategy for the next few years?

Enterprises need to stop pretending the browser session is enough and start a massive “rejiggering” of their internal and external policies. The first task is to classify every action a customer can take—reading a balance is low risk, but moving funds or closing an account requires a completely different policy for delegated automation. Second, businesses must move away from vague anti-bot clauses and create an explicit access model that defines what kinds of agents are allowed and how they must identify themselves. Third, we need better telemetry; if your security team cannot distinguish between a human and an autonomous agent, you cannot manage the risk or provide an audit trail for disputes. Finally, the most successful companies will be those that offer a “sanctioned path” by providing safe APIs or dedicated access mechanisms. If you don’t give customers a secure way to use their agents, they will continue to use credential sharing and browser automation in the shadows, leaving the enterprise with zero visibility and maximum exposure.

What is your forecast for the relationship between AI agents and enterprise workflows over the period from 2026 to 2028?

Between 2026 and 2028, we are going to see a painful but necessary evolution where the industry moves toward “controlled delegation” rather than the current binary of “allow or block.” I expect to see the rise of certified agent ecosystems and fine-grained permissions where an agent might have read-only access for spending analysis but requires extra human authentication for any transaction over a certain dollar amount. The “human vs. bot” detection models will become incredibly sophisticated, but they will be used to facilitate authorized automation rather than just stopping it. We will also see a complete overhaul of terms of service and insurance products specifically designed to bridge the liability gap when autonomous software clicks the wrong button. Ultimately, businesses that treat agent access as a competitive advantage—by making it safe and transparent—will win the loyalty of a new generation of “agent-first” consumers who no longer want to perform routine digital tasks themselves.

Do you have any advice for our readers?

The most important thing to realize is that you cannot win a war against your own customers’ desire for efficiency. If your current strategy is simply to block every agent that tries to interact with your site, you are effectively telling your most tech-savvy and high-value customers that their time isn’t important to you. Instead of viewing agents as a security incident to be mitigated, start viewing them as a new type of “power user” that requires a specific set of tools and boundaries. Build the API, define the liability early in your contracts, and create a transparent consent screen that a human can actually understand. If you provide a sanctioned, secure path for these agents to operate, you move the activity out of the shadows of credential sharing and into a governed environment where both you and your customer are protected.

Explore more

Can AI Humanize the Insurance Industry With Empathy

A legacy of cold interactions and dense administrative hurdles has long defined the insurance industry, leaving policyholders to navigate their most vulnerable moments against a wall of automated indifference. For decades, the relationship between an insurer and the insured has been characterized by a profound emotional disconnect, where individuals seeking support are met with complex legal jargon and slow, mechanical

How Can AI and CRM Architecture Drive Small Business Growth?

Navigating the complexities of the modern commercial landscape requires more than just grit and a good product because the sheer volume of digital signals generated by every customer interaction can quickly drown an unprepared entrepreneur. In the current market, the difference between a thriving enterprise and a stagnant one often comes down to how effectively a company manages its invisible

Can Recognition Solve Employee Engagement at Credit Unions?

A sincere and heartfelt expression of gratitude carries significant professional weight in a high-pressure lending environment, but it cannot offset the crushing weight of an unmanageable loan queue that grows every hour. While appreciation serves as a vital leadership tool, relying on it as a universal fix for morale risks ignoring the systemic pressures that staff face during the current

Can LeanPCS Secure Ethereum Against Future Quantum Threats?

While current Ethereum signatures are roughly 65 bytes, the quantum-resistant alternatives required for future security can reach up to 6,000 bytes. This staggering increase in data volume poses a fundamental existential threat to the scalability and economic viability of the Ethereum mainnet as the era of quantum computing approaches. If every user transaction were required to carry such a massive

AMD Mext Uses AI to Cut Server Memory Costs with SSD Storage

Modern enterprise server rooms frequently face a silent financial crisis where high-performance memory modules consume nearly ninety percent of total hardware procurement budgets. This disproportionate expenditure stems from the rigid physical limitations of traditional Dynamic Random Access Memory, which has long served as a strict gatekeeper for system performance. While storage capacities for solid-state drives have surged, the price per

Fill out the form below