Give every AI agent a named owner, job description, Maisa tells banks

Banks racing to deploy AI agents should treat each one less like a tool and more like an employee. “Each agent should have a named owner, a defined job and its own access rights — like a new hire,” Andrew Volkov , chief strategy officer at agentic AI solutions company Maisa, told FinAi News . The Madrid-based company’s “digital workers” or specialized agents each own one process and run every step as checked code. They are the auditable alternative to open-ended agents that improvise on live infrastructure, Volkov said. Larger banks are reaching the same conclusion on their own. At <a href="http://Banks racing to deploy AI agents should treat each one less like a tool and more like an employee. “Each agent should have a named owner, a defined job and its own access rights — like a new hire,” Andrew Volkov, chief strategy officer at agentic AI solutions company Maisa, told FinAi News. The Madrid-based company’s “digital workers” or specialized agents each own one process and run every step as checked code. They are the auditable alternative to open-ended agents that improvise on live infrastructure, Volkov said. Larger banks are reaching the same conclusion on their own. At Citi, a mandatory agentic development framework spans 180,000 employees across 90 countries, requiring staff to register every agent they build, Enterprise Transformation Leader Julien Courbe said in Amazon Web Services’ “Reimagine: Turning AI into Value” report published on Oct. 7. “We have inventory of the agents; we know who the owner is,” he said. Need for new models FIs that are deploying agents need a new governance framework for this novel tech, Volkov said. "An agent doesn't fit neatly into traditional risk categories," Volkov said. "It pairs a language model with system credentials and the ability to act in production," simultaneously creating exposure across model risk, operational risk, identity and access management, and third-party risk. Many FIs are creating new positions to keep tabs on rapidly evolving risk. For example, HSBC hired a global head of AI risk in enterprise risk management and Deutsche Bank is advertising a job in operational risk under its chief risk office, Volkov said. Maisa's agents that perform multiple functions within a bank such as equity research, auto loan processing and client onboarding are built to be auditable and traceable, Chief Executive David Villalon told FinAi News. “The audit trail is the execution itself,” he said, arguing that a record of the code a digital worker runs beats a transcript written afterward. “Code can be recorded, inspected and run again, which a model’s internal reasoning can’t.” Maisa has embedded a knowledge processing unit layer in all its agents. It records code the agent executed at every step, the information it looked at, what it did with it and the rationale behind each decision, Villalon said. Maisa develops its own small open-source AI models trained on bank data rather than running a frontier model, he said. “The real lock-in in AI today is depending on frontier model providers who set the price per token and can retire a model whenever they like,” Villalon said. Working with Santander Santander Bank announced in a release on Oct. 22 that it is deploying Maisa’s auditable digital workers in "hundreds of use cases." "The aim is not to automate for the sake of automating, but rather to spot where AI can simplify processes, make them run better and free up time for added-value tasks," the bank said in the release. Processes that use Maisa were not disclosed. Santander did not respond to FinAi News' request for comment."> Citi , a mandatory agentic development framework spans 180,000 employees across 90 countries, requiring staff to register every agent they build, Enterprise Transformation Leader Julien Courbe said in Amazon Web Services ’ “Reimagine: Turning AI into Value” report published on Oct. 7. “We have inventory of the agents; we know who the owner is,” he said. Need for new models FIs that are deploying agents need a new governance framework for this novel tech, Volkov said. "An agent doesn't fit neatly into traditional risk categories," Volkov said. "It pairs a language model with system credentials and the ability to act in production," simultaneously creating exposure across model risk, operational risk, identity and access management, and third-party risk. Many FIs are creating new positions to keep tabs on rapidly evolving risk. For example, HSBC hired a global head of AI risk in enterprise risk management and Deutsche Bank is advertising a job in operational risk under its chief risk office, Volkov said. Maisa's agents that perform multiple functions within a bank such as equity research, auto loan processing and client onboarding are built to be auditable and traceable, Chief Executive David Villalon told FinAi News . “The audit trail is the execution itself,” he said, arguing that a record of the code a digital worker runs beats a transcript written afterward. “Code can be recorded, inspected and run again, which a model’s internal reasoning can’t.” Maisa has embedded a knowledge processing unit layer in all its agents. It records code the agent executed at every step, the information it looked at, what it did with it and the rationale behind each decision, Villalon said. Maisa develops its own small open-source AI models trained on bank data rather than running a frontier model, he said. “The real lock-in in AI today is depending on frontier model providers who set the price per token and can retire a model whenever they like,” Villalon said. Working with Santander Santander Bank announced in a release on Oct. 22 that it is deploying Maisa’s auditable digital workers in "hundreds of use cases." [caption id="attachment_57372340" align="alignright" width="256"] (Courtesy/Bloomberg)[/caption] "The aim is not to automate for the sake of automating, but rather to spot where AI can simplify processes, make them run better and free up time for added-value tasks," the bank said in the release. Processes that use Maisa were not disclosed. Santander did not respond to FinAi News' request for comment. Register here for early-bird pricing for the FinAi Banking Summit, taking place March 1-2 in Denver. View the full event agenda here .

Give every AI agent a named owner, job description, Maisa tells banks

Santander taps Maisa to put AI agents into hundreds of processes

Nubank to expand AI underwriting into US market

Related Posts

Nubank to expand AI underwriting into US market

Banks on the hook for agents’ actions, OCC says

Nearly 85% of TD’s 2026 patents involve AI

EMERGING FINTECH DIRECTORY

FinAi Podcast

SPONSORED

The Next Generation of Cyberthreats: Powered by AI

Build an Antifragile Strategy to Outperform the Market

How AI and Product Experts Turn Fuzzy Requirements Into Focused Dev-ready Roadmaps

Connect

Welcome Back!

Retrieve your password

Continue Reading with FinAi News Premium - Less than $2/Day

Give every AI agent a named owner, job description, Maisa tells banks

Santander taps Maisa to put AI agents into hundreds of processes

Nubank to expand AI underwriting into US market

Related Posts

FinAi Podcast

Connect

Welcome Back!

Retrieve your password

Unlock This Article

Continue Reading with FinAi News Premium - Less than $2/Day