BNY and Google are teaming up to supercharge the bank's AI ambitions with Gemini 3

• BNY is adding Google's Gemini 3 to its enterprise AI platform Eliza, the bank said Monday.
• The move focuses on "agentic" AI that supports multi-step financial workflows.
• It comes as Wall Street's AI arms race intensifies among Wall Street banks and investment firms.

The race on Wall Street to deploy AI agents capable of handling end-to-end tasks is intensifying.

On Monday, BNY said it will embed Google Cloud's agentic AI technology — including Google's latest model, Gemini 3 — into the bank's internal AI platform Eliza, named after the wife of bank founder Alexander Hamilton. The upgrade of its technology stack is the latest step the bank — one of New York's oldest operating businesses — is taking to advance its AI ambitions.

By integrating Google Cloud's tech into Eliza — which already draws on multiple large language models — the firm is betting employees will be able to move faster through daily tasks, Sarthak Pattanaik, the bank's chief data and AI officer, told Business Insider. Take the tedious manual work of client onboarding: staff juggle document collection, verify ID and tax forms, locate key details, look up risk information, and log everything into internal systems. He said agentic AI could help orchestrate those steps, break tasks into smaller components, and deliver the required information in a more streamlined flow.

"Eliza, through the agent tech workflow, is able to make the process much more simpler, efficient, and orchestrated, much more seamlessly," Pattanaik said.

Gemini 3, which debuted in mid-November, can interpret text, images, tables, PDFs, and audio together, allowing employees to load mixed financial materials and have the model interpret and synthesize what matters.

BNY's generative AI build-out began accelerating in 2023, and the bank has been keen to position itself as an early adopter of AI. Eliza now supports more than 120 automated tasks. Pattanaik added that nearly the entire firm has completed generative AI and responsible AI training.

In the bank's third-quarter earnings call, BNY CEO Robin Vince told analysts that the bank was leveraging agentic AI to deploy over 100 "digital employees" that are working "side-by-side" with staff on tasks such as payment validations and code repairs.

"We believe our AI opportunity is significant, and we are pursuing it with urgency," said Vince.

Earlier this year, it announced a similar partnership with OpenAI, the maker of the popular consumer-focused large language model ChatGPT. On its website, it says it was "the first major bank to deploy an AI Supercomputer (powered by NVIDIA) to accelerate processing capacity."

Safety protocols

While exciting, its expansion into agentic systems raises questions about how highly regulated institutions like BNY will ensure data privacy. Both BNY and Google emphasized that deploying agentic AI inside a regulated institution requires boundaries around what the technology can see, decide, or escalate.

Pattanaik said deploying agentic AI inside a bank demands strict oversight. He said each agent must pass an internal model-risk review before it goes live and emphasized that the systems are governed by tight access controls that determine what information they are allowed to use.

Once deployed, he added, the team monitors the agent's performance daily and incorporates those results into a continuous feedback loop.

Rohit Bhat, Google Cloud's head of financial services, told Business Insider that Google provides the safety mechanisms that restrict how agents communicate and what data each one can access. Agents have "development kits" and "protocols" to govern their communication with one another, Bhat said.

"The primary purpose of kits and protocols," he added, "is to establish that communication pathway with those boundary conditions of, 'I'm only allowed to talk to this agent for the set of reasons and nothing else.'"

Why Google thinks it can help Wall Street

Major banks are embracing a mix of homegrown and external AI tools. Goldman Sachs has expanded its internal platforms while experimenting with technology from startups like Cognition Labs.

Dealmakers and investors have adopted tools from newer entrants like Hebbia, which offers prompt libraries and deep search. Morgan Stanley has deployed OpenAI technology to assist its financial advisors, while executives at JPMorgan have spoken publicly about the potential for junior employees to manage teams of AI agents.

Bhat said financial firms are becoming an important testing ground for agentic AI because their workflows involve heavy documentation, strict rules, and significant risk management.

Google's pitch is that Gemini can reason through lengthy materials while staying anchored to a firm's internal policies — a prerequisite for deploying agents across custody, markets, or onboarding.

"You need to make sure that whatever these agents are doing is grounded in the business context and in the business specificity," he said. "That requires these models to be able to understand, and then also adhere to, certain policies and the rules," to make sure they meet "the standards that you would want them to."