Agentic AI in banking refers to AI systems that can pursue user-defined financial goals, make context-aware decisions within approved limits, and execute multi-step tasks with minimal human intervention.
Banking UX has always been about helping people do things – send money, pay bills, apply for a loan, or open an account. Agentic AI flips that model. Now the AI can take action on the customer's behalf. That means great UX is now about helping people understand what happened, why it happened, and giving them confidence that they're still in control.
Imagine you wake up, open your banking app, and see that your AI assistant has automatically moved your idle cash into a higher-yield savings product you previously authorised. The decision aligns with your financial goals and follows the limits you set.
Do you feel reassured? Or do you instinctively wonder why it happened, whether it was the right decision, and if you can reverse it?
That moment captures the real challenge of agentic AI in banking. Few markets illustrate this shift more clearly than the United Arab Emirates. Dubai has launched a two-year initiative to accelerate Agentic AI adoption across its private sector, while the UAE Government aims to transition 50% of federal government services and operations to autonomous AI by 2028.
This guide explores what agentic AI in banking is and the UX principles needed to design transparent and trustworthy banking experiences. It also examines the opportunities and design considerations specific to Dubai and the wider UAE. So, let's get started.
What Is Agentic AI in Banking UX?
Agentic AI in banking UX is the design of banking experiences where AI can understand a customer's goals, make context-aware decisions, and carry out approved actions while keeping people informed and in control. Unlike traditional AI assistants that simply answer questions or respond to prompts, agentic AI can plan and execute multi-step tasks within customer permissions, organisational policies, and regulatory guardrails.
The defining characteristic here is agency, you see. Instead of waiting for instructions at every step, an AI agent can interpret an objective, evaluate relevant information, coordinate actions across connected banking systems, and decide when to act independently, seek customer confirmation, or hand control back to a person.
The difference becomes clearer in a real banking scenario. A chatbot can tell you your account balance or explain your loan eligibility. An agentic AI system, on the other hand, can detect that your account is likely to fall below your preferred balance before payday, transfer funds based on rules you've already approved, and clearly explain every action it has taken. Rather than simply responding to a request, it works proactively towards a financial outcome you've already authorised.
Therefore, the challenge for designers is no longer helping users complete a task. It's also about designing experiences that make autonomous actions easy to understand, explain, and control.
Also Read: Agentic AI UX Design - 5 UX Patterns That Work
Agentic AI vs Traditional AI, Chatbots and RPA: What are the Differences?
The clearest way to understand agentic AI is to place it beside the tools it is often confused with.
| Type of system |
What it does |
How work gets done |
| Rule-based chatbot |
Answers predefined questions using fixed rules |
You complete the task |
| Generative AI |
Creates content, summaries, and recommendations from prompts |
You act on its output |
| Robotic Process Automation (RPA) |
Automates repetitive, rule-based workflows |
It follows predefined rules |
| Agentic AI |
Pursues goals, reasons, and executes approved actions |
It acts within your approved limits |
The jump from generative AI to agentic AI is the jump from recommendation to execution. A generative assistant suggests the next step. But it's an agent that can carry it out within approved limits.
Common Agentic AI Architectures in Banking
Banks typically deploy agentic AI in three ways.
Task agents focus on a single goal, such as monitoring cash balances, detecting unusual transactions, or automating recurring payment workflows.
Workflow agents coordinate multiple steps within a business process, such as customer onboarding, KYC verification, or loan processing.
Multi-agent systems bring together several specialist agents – one analysing credit risk, another verifying customer information, and another checking regulatory compliance – to collaborate on more complex decisions.
This collaborative approach is already emerging in financial services. Moody's, for example, describes agentic workflows built around specialist AI agents that work together on tasks such as credit assessment, compliance, and KYC before producing auditable and explainable outputs.
Also Read: Top 10 Fintech UX Design Practices Every Team Needs in 2026
How Is Agentic AI Changing Digital Banking?
Research from Capgemini estimates that agentic AI could generate up to US$450 billion in additional value across the banking industry by 2028 through new revenue opportunities and operational efficiencies.
The momentum is equally visible across the Gulf. A 2026 Roland Berger survey found that almost four in five organisations had already embedded AI into their long-term business strategy. Those numbers point to something bigger than technology adoption. They suggest that banks increasingly see autonomous AI as part of their future operating model, and not just another digital feature.
The story of digital banking has always been about giving customers better tools.
First came internet banking, making it possible to manage money without visiting a branch. Then mobile apps put everyday banking into our pockets. Over time, features like biometric login, instant payments, spending insights, and personalised recommendations made the experience faster and more convenient.
Through all of those changes, however, one thing stayed the same. Customers remained responsible for every decision and every action.
Agentic AI changes that relationship. Instead of simply helping someone decide what to do next, an AI agent can monitor situations, coordinate multiple banking services, and carry out approved actions on the customer's behalf. The customer still decides the goals, permissions, and boundaries. The AI takes care of the routine execution.
That may sound like a subtle change, but it fundamentally reshapes banking UX. The interface is no longer just where people complete tasks. It's where they understand what the AI is doing, why it made a decision, and whether they want to let it continue.
Also Read: How to Design Agentic AI Systems Users Can Trust
Why UX Will Matter More in Agentic Banking?
Building an intelligent AI agent is only half the challenge. The bigger question is whether customers will trust it enough to let it act on their behalf.
Capgemini's research highlights why that matters. Trust in fully autonomous AI agents fell from 43% to 27% over the course of a year, reflecting growing caution as organisations moved from experimentation towards real-world deployment.
People lose trust when they don't understand what it's doing or feel they've lost control.
When an AI recommends a better savings product, many customers are comfortable. But when it automatically moves money, even within approved limits, the questions change immediately.
- Why did it do that?
- Can I undo it?
- Can I stop it from happening again?
- What happens next?
Well, these questions are about the experience.
So yes, the banks that earn lasting customer trust will be the ones that make every autonomous decision easy to understand, straightforward to review, and always under the customer's control.
That's why the future of agentic banking isn't defined by AI alone. It's defined by the experience surrounding it.
Why Is the UAE Leading Agentic AI Adoption?
The UAE appointed the world’s first Minister of State for Artificial Intelligence in 2017 and launched the UAE AI Strategy 2031, signalling a long-term commitment to AI-driven innovation across government and industry.
More recently, Dubai has introduced initiatives to accelerate Agentic AI adoption across the private sector, encouraging organisations to rethink how autonomous AI can improve services and productivity.
Regional events such as the Middle East Banking Innovation Summit (MEBIS) reflect a growing industry focus on AI governance, automation, customer trust, and the future of autonomous banking.
For banks operating in Dubai and the wider UAE, agentic AI is no longer something to watch from the sidelines. It is becoming a strategic capability that will influence everything from customer experience and operational efficiency to compliance and long-term competitiveness.
The question, therefore, is how to design experiences that make customers comfortable enough to let it act on their behalf.
How UX Changes as AI Becomes More Autonomous
The more autonomous the system becomes, the more people expect transparency, explainability, and control.
| AI capability |
What it does |
Customer's role |
UX priority |
| Answers questions |
Provides information on request |
Makes every decision |
Clear, accurate responses |
| Recommends actions |
Suggests personalised next steps |
Reviews and decides |
Explainability and comparison |
| Executes approved actions |
Carries out authorised tasks within set limits |
Sets permissions and supervises |
Visibility and easy overrides |
| Coordinates workflows |
Manages connected banking tasks across systems |
Sets goals and intervenes when needed |
Progress tracking and audit trails |
| Continuously optimises |
Monitors goals and adapts within agreed boundaries |
Defines objectives and governance |
Transparency and instant intervention |
How Much Human Oversight Should Agentic AI Have?
Well, not every banking task requires the same level of human involvement. The right level of oversight depends on the potential impact of the AI's decision.
For low-risk activities, such as categorising transactions or generating spending insights, AI can often operate with minimal supervision. As decisions become more sensitive, say approving loans, detecting fraud, or managing investments, human oversight becomes increasingly important.
AI governance frameworks commonly describe three levels of human oversight:
| Oversight model |
How it works |
Typical banking use case |
UX priority |
| Human in the loop |
A person reviews and approves every action before it happens. |
Loan approvals, large-value transactions |
Clear, accurate responses |
| Human on the loop |
The AI acts within approved limits while a person monitors and can intervene. |
Fraud monitoring, payment automation |
Explainability and comparison |
| Human out of the loop |
The AI operates autonomously within predefined rules and governance controls. |
Low-risk back-office processes, routine operations |
Visibility and easy overrides |
The greater the potential impact of an AI decision, the greater the need for transparency, approvals, explainability, and easy intervention.
Core UX Design Principles for Agentic Banking
Six principles separate agentic banking experiences people trust from ones they abandon. These include:
1. Show Intent Before Action
Where appropriate, give customers a preview of what the agent plans to do, why it intends to do it, and the likely impact before it acts. Pre-action visibility helps customers understand and validate an agent’s behaviour before it executes. By contrast, post-action notifications explain decisions only after they have already taken effect.
2. Design Granular Controls
People rarely trust an AI agent equally across every financial task. Someone may be comfortable letting it pay recurring bills automatically but still want to approve every investment transaction. Designing autonomy as a single on/off setting overlooks these different trust thresholds. Instead, let customers define guardrails based on transaction type, amount, risk level, and timing, and allow those permissions to evolve as confidence grows. For example, a customer might allow automatic utility bill payments while requiring approval for any equity trade above AED 5,000.
3. Explain Decisions in Context
Simply telling a customer that "the AI analysed hundreds of variables" explains very little. Connecting the decision to the customer's own goal explains far more: "You asked to keep 20% of your money easily accessible while earning more on the rest. Moving AED 50,000 into a short-term investment helps achieve that balance." Context transforms technical reasoning into an explanation people can understand, evaluate, and trust.
4. Make Human Intervention Effortless
Customers should be able to pause, question, or reverse AI-initiated actions wherever operationally possible. Friction in the intervention path is more than a usability issue as it can undermine confidence in autonomous systems. In banking, where financial outcomes are at stake, providing clear intervention and escalation paths becomes even more important.
5. Let Trust be Earned
Customers rarely trust an agent because it promises good outcomes. They build trust as it consistently demonstrates value over time. Show that history: "Over the past month, your agent helped you avoid AED 1,200 in overdraft fees and optimise your savings based on the preferences you set—here's how." Demonstrating past performance and aligning actions with customer goals builds confidence more effectively than reassurance hidden in terms and conditions.
6. Make Every Decision Traceable
Customers should be able to ask, at any time, "Why did you move that money last Tuesday?" and receive a clear, plain-language explanation. A persistent decision history that customers can revisit, question, and understand is essential for making long-term delegation feel transparent rather than opaque.
Agentic AI Use Cases in Banking
Agentic AI is being adopted across banking for tasks that require persistent monitoring, decision-making, and coordinated action. These range from everyday retail banking journeys to complex institutional operations where speed, scale, and adaptability matter.
1. Retail and Personal Banking
In day-to-day banking, agents handle recurring payments, move money between accounts to avoid overdrafts, and shift idle cash into customer-approved savings or investment products. These are often among the first use cases explored because customers can define clear rules and spending limits, making automation easier to supervise.
2. Wealth and Portfolio Management
For savers and investors, agents can monitor markets around the clock and recommend or execute customer-authorised portfolio rebalancing to keep it aligned with a stated goal, such as "grow steadily but keep a fifth of my money easy to access." They can flag when a holding drifts too far from target. Because the sums are larger, these agents usually keep a human on or in the loop for bigger moves.
3. Credit and Lending
Agents can support credit assessment and accelerate lending decisions by drawing on a wider range of data than a traditional score. This speeds up decisions dramatically. In many jurisdictions, including the UAE, high-impact lending decisions are expected to include appropriate human oversight and clear mechanisms for review or appeal.
4. Fraud, Compliance and KYC Monitoring
Some of the most mature applications come from the back office. Agents screen transactions in real time, flag suspicious patterns before or shortly after suspicious transactions occur, run KYC and anti-money-laundering checks, and validate identity documents. These use cases are generally more mature because AI supports detection, monitoring, and investigation rather than independently making customer financial decisions.
5. Corporate and Treasury Operations
For business clients, agents optimise cash flow, manage treasury positions and handle routine currency operations. The efficiency case here is huge, but so is the need for oversight, because a single error can cascade across large sums and multiple systems.
6. Multi-agent Orchestration Behind the Scenes
Many advanced agentic AI systems rely on multiple specialised agents rather than a single one. Moody’s has described using specialist agents for tasks such as financial risk analysis, macroeconomic assessment, and credit evaluation. Their outputs are then validated, combined, or reviewed through orchestration and quality assurance mechanisms before producing a final recommendation. To the customer, this complexity remains invisible. They experience a single, coherent response, even though multiple agents may have contributed behind the scenes.
7. Customer Service and Relationship Management
Customer service is one of the fastest-growing applications of agentic AI in banking. Rather than answering isolated questions, AI agents can manage end-to-end service journeys, such as opening accounts, replacing cards, tracking disputes, updating customer information, or scheduling appointments by coordinating across multiple systems. When a request becomes complex or requires judgement, the agent can transfer the conversation to a human adviser with the relevant context, reducing friction while maintaining continuity of service.
Real-World Case Studies: Agentic and AI-Driven Banking
Let’s understand how financial institutions are beginning to apply autonomous AI capabilities across customer service, operational workflows, and investment research.
ANZ Bank
Australia's ANZ Bank is gradually introducing agentic AI into its business banking operations through an AI-powered CRM built on Salesforce Agentforce. The platform:
- Automates routine and multi-step workflows
- Brings together data from multiple banking systems
- Helps bankers access customer insights more efficiently
BlackRock's Asimov
BlackRock has developed an internal AI platform called Asimov to support investment professionals with research and market analysis. Rather than making investment decisions independently, the platform helps analysts process large volumes of financial information, identify relevant insights, and accelerate research.
Agentic Banking in Dubai and the UAE: A Regulatory-First Approach
The UAE has emerged as one of the world’s leading markets for AI adoption. Backed by the UAE Strategy for Artificial Intelligence 2031 and guidance from the Central Bank of the UAE, banks are encouraged to adopt AI while maintaining transparency, accountability, and consumer protection.
Why Dubai Is Becoming a Testbed for Agentic Banking
In 2026, Sheikh Hamdan bin Mohammed launched a two-year initiative to accelerate the adoption of Agentic AI across Dubai's private sector, supported by training programmes, incubators, and dedicated funding through Dubai Chambers.
At the national level, the UAE Government has also announced a framework to transition 50% of government sectors, services, and operations to Agentic AI within two years.
Therefore, for banks, this creates a unique environment where innovation is being encouraged alongside governance. That combination makes Dubai one of the most closely watched markets for designing and deploying trustworthy agentic banking experiences.
The UAE Regulatory Stack Every Product Leader Must Design Around
Agentic banking in the UAE has to be designed around a layered set of rules:
- CBUAE Guidance Note (February 2026): The Central Bank of the UAE issued guidance on the responsible use of AI and machine learning by licensed financial institutions. It states that human-out-of-the-loop systems should be used only for low-risk, non-material processes with appropriate safeguards. It also makes clear that licensed financial institutions remain accountable for AI systems, including those provided by third-party vendors.
- DIFC Regulation 10: The Dubai International Financial Centre introduced Regulation No. 10 in 2023 to update its data protection framework for autonomous and semi-autonomous systems. It became fully enforceable on 1 January 2026 and is widely regarded as one of the region's earliest AI-focused regulatory measures.
- ADGM sandbox: Abu Dhabi Global Market operates a regulatory sandbox that allows financial institutions to test innovative technologies under regulatory supervision before wider deployment.
- UAE PDPL: The federal Personal Data Protection Law governs how personal data is collected, processed, stored, protected, and transferred, providing the foundation for responsible customer data management.
Which Banks in Dubai and the UAE Have Adopted Agentic AI?
The UAE has some of the most AI-mature banks in the region, but documented agentic AI deployments are still nascent. The inaugural Evident AI Index for Banks – Middle East & Africa, issued in June 2026, put three banks in the UAE among the ten most advanced in terms of AI in the region. The index is a measure of total AI capabilities, not agentic AI adoption, but it is a valuable baseline for identifying those institutions driving the next stage of AI in banking.
First Abu Dhabi Bank (FAB) offers one of the clearest public examples of agentic AI in the UAE. The Evident AI Index highlights documented agentic AI use cases across payment processing, relationship management, and developer support, while also recognising FAB's strong focus on building AI capabilities at scale. FAB ranked third in the inaugural regional index.
Emirates NBD, Dubai's largest bank, topped the Evident AI Index and was the only institution to rank in the top three across multiple assessment pillars. The bank has deployed more than 50 AI use cases across its operations and launched an Acceleration-to-Enterprise partnership with Techstars to support AI and fintech innovation, including emerging agentic finance applications. It also participated in one of the region's first publicly demonstrated voice-enabled agentic payment experiences, where an AI agent completed a customer-authorised service charge payment using an Emirates NBD Darna Visa card.
Mashreq ranked tenth in the Evident AI Index, reflecting its broader AI maturity. Through Mashreq Neo, its Emma virtual assistant, and partnerships such as Visa and Rezolve AI, the bank is expanding AI-powered customer experiences while exploring agentic commerce capabilities. Much of its progress in agentic AI is currently being driven through strategic partnerships rather than publicly documented in-house deployments.
Beyond the top ten, Abu Dhabi Commercial Bank (ADCB) and Dubai Islamic Bank (DIB) also featured in the Evident AI Index's regional top 25. Their inclusion reflects strong AI maturity, although public evidence of agentic AI deployments remains limited as of mid-2026.
What Does a Dubai Deployment Roadmap Look Like?
For banks in the UAE, the most practical approach to agentic AI is to start small and scale responsibly.
- Begin with a well-defined, lower-risk use case where outcomes can be measured, and governance is easier to maintain.
- Design the guardrails first. That is, customer controls, clear disclosures, human oversight, and intervention mechanisms, before expanding the agent’s authority.
- Build compliance reviews into every stage of development, aligning with CBUAE requirements and, where applicable, the regulatory frameworks of the DIFC or ADGM.
- Pilot the solution under close supervision, measure customer trust alongside operational outcomes, and use those insights to refine the experience before wider deployment.
How to Design and Implement Agentic AI in Banking UX
While implementing agentic AI in banking, it is important to treat trust as the primary metric and build the experience outward from the customer's need for control. The steps below turn the principles in this guide into a practical sequence.
- Start with an experience audit: Before adding autonomy, understand the friction in your current journeys. A structured UX audit maps where customers hesitate, drop off, or lose confidence, so that agents are aimed at real problems rather than bolted on for novelty.
- Map autonomy to trust: Match each use case to an appropriate level of autonomy and the right amount of human oversight. Low-risk, high-frequency tasks can run more freely while high-impact decisions keep a human in or on the loop.
- Build the control layers first: Design the guardrails, previews, override, and audit views before you expand what the agent can do. Without clear customer controls, even technically capable AI systems are unlikely to earn long-term user trust.
- Govern for auditability from day one: Document how the agent reaches decisions, keep clear records, and validate outputs regularly. Moody's stresses that transparency and auditability are essential in regulated finance. And UAE regulatory guidance similarly emphasises transparency, governance, accountability, and appropriate human oversight for AI systems.
- Measure trust, not just efficiency: High opt-out or override rates often indicate that customers do not yet trust the experience, regardless of operational efficiency gains.
Planning Your Agentic Banking Journey?
Agentic AI has the potential to transform banking, but its success will be measured by how confidently customers use it and how responsibly banks govern it.
At Onething Design, we've spent years helping financial institutions simplify complex journeys and build digital experiences that customers trust. Our work with organisations such as Kotak Mahindra Bank, RBL Bank, and HDFC Securities has given us first-hand experience designing for highly regulated environments where usability, transparency, accessibility, and compliance must work together.
With agentic AI becoming the next frontier of digital banking, we believe the biggest opportunity lies in designing better experiences around them.
If you’re exploring agentic AI for your bank – whether you're validating the opportunity, redesigning customer journeys, or preparing for rollout – we'd be happy to help. Get in touch with our team to discuss how thoughtful UX and Agentic Experience Design (AXD) can help you build AI-powered banking experiences your customers will trust from day one.