Join our daily and weekly newsletters to get the latest updates and exclusive content regarding the leading scope of artificial intelligence. Learn more
The financial services industry is one of the most regulated sectors. It also manages huge amounts of data. Aware of caution, financial companies have slowly added AI and AI generative agents to their stables.
The industry is not foreign to automation. But the utilize of the term “agent” has been muted. And, understandable, many in the industry took Very cautious attitude towards AI generativeEspecially in the absence of regulatory framework. Now Banks such as JP Morgan AND Bank of America debuted AI powered assistants.
The bank at the head of the trend is Bins. Financial Services The company founded by Alexander Hamilton updates its AI tool, Eliza (named after Hamilton’s wife), developing them in the resources of the agent many times. The bank perceives AI agents as valuable assist to its sales representatives, while more involving its clients.
Multi -agative approach
Sarthak Pattanaik, head of Bny’s Artificial Intelligence Hub said Venturebeat in an interview that the bank began to determine how to combine many units so that it was uncomplicated to get their information.
Bny created the main agent of recommendations for his various teams. But it did more. In fact, he uses an agent architecture many times to assist his sales team to issue appropriate recommendations for customers.
“We have an agent who has everything [the sales team] know[s] About our client – said Pattanaik. “We have another agent who talks about products, all products that the bank has … from liquidity to security, payment, treasure and so on. Ultimately … We try to solve the client’s need through the possibilities we have, product capabilities. “
Pattanaik added that his agents have reduced the number of people with whom many employees dealing with clients must talk to determine a good recommendation for clients. So “instead of sellers talking to 10 different product managers, 10 different customers, 10 different segmental people, all this is now done through this agent.”
The agent allows the sales team to answer very specific questions that customers may have. For example, does the bank support foreign currencies such as Malaysian Ringgit, if the customer wants to launch a credit card in this country?
As they built it
Possibilities of recommending many agents debuted in the tool Eliza Eliza Bny.
There are about 13 agents who “negotiate with each other” to find a good product recommendation, depending on the marketing segment. Pattanaik explained that funds include functional factors such as customer factors, as well as segmental factors that affect structural and unstructured data. Many agents in Eliza have a “sense of reasoning”.
The bank understands that the agent’s ecosystem is not fully agency. As Pattanaik noted: “A fully agency version would be that it would automatically generate PowerPoint, which we can give to the client, but we don’t do that.”
Pattanaik said that the bank turned to the Microsoft autogen to revive its AI agents.
“We started with autogen because it is open source,” he said. “We are basically a construction company; Wherever we can use Open Source, we do it. “
Pattanaik said that AutoGen provided the bank with a set of solid handrails that he can utilize to ground many agents’ answers and make them more deterministic. The bank also looked at Langchain to architect the system.
Bny built a frame around the agency system, which gives agents a plan to answer demands. To achieve this, the company’s AI engineers worked closely with other bank departments. Pattanaik emphasized that BNY has been building critical platforms for years and scaled products such as platforms and security platforms. This deep knowledge bench was the key to helping AI engineers responsible for the agent platform in providing agents to specialized specialist knowledge.
“Having a smaller hallucination is a feature that always helps, compared to having AI engineers running the engine,” said Pattanaik. “Our AI engineers cooperated very closely with whole -grain engineers who built critical systems of mission to help us justify the problem. It’s about componenting to make it multiple. ”
For example, building the main agent of procurement in this way allows it to be developed by various BNY business lines. It acts as a micros service “which still learns, reasonable and working”.
Expansion of Eliza
As the agency rates develop, BNY plans to further improve its flagship AI, Eliza tool. Bny issued a tool in 2024, although it has been in development since 2023. Eliza allows BNY employees to access the AI application market, obtain approved data sets and look for information.
Pattanaik said that Eliza already provides a plan on how to go ahead with AI agents and offer users a more advanced, bright service. But the bank does not want to be in stagnation and wants Eliza’s next iteration to be more bright.
“What we built with Eliza 1.0 is a representation and aspect of learning things,” said Pattanaik. “Thanks to 2.0 we will improve this process, and also ask how we will build a great agent? If you think about agents, it’s about something that you can learn and reason, and at one point to provide some actions, what a break is, this is not a break and so on. This is a direction in which we will build 2.0, because many things should be established in terms of risk handrail, explanatory ability, transparency, connections and so on before we become completely autonomous. ”