Ever faced frustrating wait times with your bank’s customer service or wished for 24/7, personalized financial support? The solution lies in the next innovative wave: Agentic AI.
Agentic AI represents a leap in technology—an autonomous system that not only sets goals but also analyzes, decides, and acts on its own.
Unlike traditional AI, it learns in real-time, adapting dynamically to tackle complex decisions and deliver seamless support.
We recently connected with Amdocs–a digital transformation partner for banks worldwide–to explore topics such as Agentic AI and how it enhances customer experience while addressing potential concerns.
Of course, as with any transformative innovation, there are challenges, skepticism, and ethical considerations.
However, banks that embrace Agentic AI early stand to gain a competitive edge, while those that hesitate risk being left behind.
Ramya Patel, Data & Intelligent Analytics Leader, Americas at Amdocs, has shared key insights on this transformative trend.
Our chat with Ramya Patel is shared below.
Crowdfund Insider: How would you define Agentic AI?
Ramya Patel: GenAI Agents are intelligent AI systems designed to handle tasks on their own, like creating content, making decisions, and taking action. They blend the power of large language models with specialized skills and automations, allowing them to carry out tasks, offer personalized recommendations, and manage interactions and processes from start to finish.
Crowdfund Insider: How does it differ from other types of AI?
Ramya Patel: The big difference with Agentic AI is that it doesn’t just process information—it takes action. Unlike traditional AI, which typically follows set rules or waits for human input, Agentic AI can set its own goals, analyze situations, and make decisions on its own. It also learns in real time, so it can adapt and respond to complex, changing situations without needing constant guidance.
For example, if a customer calls the bank wanting a car loan, gen AI could provide them with options and next steps whereas agentic AI would make recommendations based on looking at the person’s financial picture and lead them through the entire process of securing a loan.
Crowdfund Insider: What are the benefits of Agentic AI?
Ramya Patel: Agentic AI is going to revolutionize customer engagement, especially in industries like banking where customers are dealing with one of the most sensitive topics there is – money – and want 24/7 access to support. These AI systems drive higher productivity and enable banks to stay ahead of the competition in an increasingly dynamic market.
For banking customers, quick decision-making and fast action are super important for optimizing financial outcomes. Delays can lead to missed opportunities or financial inefficiencies.
Agentic AI addresses this by leveraging data-driven insights to provide real-time, personalized offers and proposals. It not only analyzes a customer’s financial behavior and preferences but also enables instant execution of recommended actions, ensuring a seamless transition from decision to action.
For example, if a customer receives a loan pre-approval based on their transaction history and credit score, Agentic AI allows them to accept the offer and process the disbursement instantly. Similarly, if it identifies an opportunity to optimize savings or investments, it can suggest a strategy and execute the transaction upon approval.
Customer dissatisfaction statistics further underscore the urgency of transformation and implementation of Agentic AI. According to Amdocs’ research, 39% of people are dissatisfied with the wait times for human agents and 38% are frustrated with their limited availability during standard business hours. Being able to offer 24/7 support would be invaluable in solving these customer pain points.
And it’s not only the availability of human agents that aren’t satisfactory, 40% report dissatisfaction with the empathy and politeness demonstrated by human agents during support interactions. AI agents will enter every customer interaction with the data of past interactions and have the ability to be personalized to the customers’ preferences, such as language, gender, tone, etc.
Crowdfund Insider: What are the possible drawbacks?
Ramya Patel: Agentic AI will play a big role in tackling customer experience issues, but there are still some concerns customers have about using it.
Amdocs’ research shows that 60% of customers worry that AI agents might misinterpret what they say. Privacy and transparency are also major concerns—50% of customers fear their data might be used for purposes they’re unaware of, and 46% are unsure when they’re interacting with a human versus a GenAI agent.
Bias in decision making, privacy and security concerns, regulatory ramifications, are some of the top concerns and a challenge as it leads to unfair outcomes.
Another challenge is the “uncanny valley,” that eerie feeling people get when robots or virtual simulations look human-like but aren’t quite convincing enough. It’s a bit like the negative reaction we saw to The Polar Express, where people found the characters’ lifelike appearance unsettling because something about them didn’t seem quite right.
A way around AI representation falling into this category is to tailor the portrayal of the AI to the target audiences’ preferences and comfortability. For example, in Japan, AI developers have found that people respond best to AI when it is in the form of Manga characters.
Crowdfund Insider: Who are the most likely adopters?
Ramya Patel: The most likely adopters of Agentic AI technology will be companies at the forefront of innovation, particularly in industries like finance. Financial institutions in Europe are already paving the way here.
For example, banks are embracing new technologies, although carefully. Banks who want to remain competitive will begin to collaborate more with fintech companies who are pioneering Agentic AI. These partnerships will be key in driving adoption, as both industries combine their strengths to harness the full potential of Agentic AI and enhance their services. This merging of industries will allow them to leverage the technology more effectively, creating new opportunities for growth and efficiency.
Crowdfund Insider: What’s the best way to get started with Agentic AI?
Ramya Patel: The best way to implement Agentic AI is through a phased approach that prioritizes customer comfort and trust.
Start by introducing it gradually, giving customers the option to interact with human agents if they prefer. As I mentioned earlier, many people are frustrated by the long wait times for human agents. AI agents can be offered as an alternative during peak times, while ensuring customers retain their place in the queue if they choose to wait for a human.
Privacy and transparency are critical. Think back to the early days of e-commerce—it’s important to clearly communicate how data is being used or processed, let customers know if calls are being recorded, and always be upfront when they’re interacting with an AI agent. Reassure them that their interactions are private and secure to build trust.
Make the capabilities of the GenAI agents clear to customers, explain how the “agent” can handle complex queries, understand natural language, and personalize interactions. Rebrand as far away as possible from “chatbots” and instead position these GenAI agents as sophisticated tools that can elevate customer interactions. Personalization is key—showcase how Agentic AI agents can adapt to language preferences or specific communication styles to provide a customized experience.
Finally, stay well away from the “uncanny valley.” If your GenAI agents come across as too human-like, it could feel unsettling for customers. Focus on creating AI interactions that feel intuitive, helpful, and authentic without crossing into uncomfortable territory.
Crowdfund Insider: Is there anything else you would like to add?
Ramya Patel: One important point to emphasize is the crucial role that data quality plays in successfully implementing AI. Getting data properly prepared for GenAI is key to achieving accurate, reliable results. If the bad data is fed into Agentic AI systems, the outcome will have compound flaws.
For example, a Fintech company in the APAC region discovered that their data was biased because it only included information from male users, under the false assumption that men were the primary decision-makers in household financial matters.
This caused skewed, biased results. To correct this, they had to generate new, synthetic datasets that included data reflecting female perspectives, ensuring their AI systems produced fair and balanced outcomes.
This example underscores the importance of addressing both data quality and diversity in order to avoid biases and enable AI to deliver meaningful, unbiased insights.