Bank of America uses ‘Orchestra’ tool to decide if task needs open-source or closed-source model

Bank of America is taking a pragmatic approach to AI by routing tasks to specific models based on a task’s complexity. Depending on the intricacy and computational needs of the task, the bank’s in-house built AI tool called ‘Orchestra’ chooses which model — open-source, proprietary LLM or closed-source — is deployed, Hari Gopalkrishnan , chief technology and information officer, said during the Fortune AIQ event on Oct. 1. [caption id="attachment_57380518" align="alignright" width="255"] (Courtesy/FinAi News)[/caption] For basic classification work, the system routes requests to open-source, open-weight models running on the bank's GPUs, Gopalkrishnan said. The bank did not disclose the GPU provider or how many it has. It directs more complex reasoning tasks to proprietary models, he said, adding that this flexibility helps manage operational expenses. "We have a model router philosophy, where we essentially want to leverage the best of what the world has to offer," Gopalkrishnan said. "There are tremendous innovations happening in Silicon Valley and other places. There's also a lot of good open-weight models that exist." AI savings The bank's broader AI strategy centers on Erica, its conversational AI platform launched roughly eight years ago, Gopalkrishnan said. Erica now processes 3.6 billion transactions annually . The platform has had a material effect on the bank's workforce: Without it, that many customer service calls would require an additional 11,000 employees in call centers, he said. Bank of America starts every project by identifying a customer’s problem, not by asking whether AI is the right solution, Gopalkrishnan said. Sometimes deterministic models or simpler technologies prove more appropriate. The bank also stressed the importance of guardrails and human-in-the-loop decision-making, particularly for high-stakes applications. It has declined to rush toward fully autonomous systems, instead focusing on "assistive agents" that work alongside human employees. "We're not in a rush to create wildly autonomous systems that are just running wild with goals and planning," the executive said. "There is so much juice to be squeezed right now with assistive agents that are actually working with humans in loop." FinAi Lending Summit, set for Oct. 7-8 in Las Vegas, will include speakers from Fifth Third and Capital One as well as a fireside chat with U.S. Bank Senior Vice President – Lending Operations Group Manager Suzanne Rathbun . To learn more about the 2026 event and register, visit here .

Bank of America uses ‘Orchestra’ tool to decide if task needs open-source or closed-source model

Bank deferred hiring 11K call center employees thanks to Erica, exec says

Why Flagstar’s AI agents don’t have access to internet

Related Posts

Why Flagstar’s AI agents don’t have access to internet

Plaid partners with four LLMs, but AI still lags on financial advice

AI writes 60% of code at BNY, up 20 percentage points QoQ

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

Bank of America uses ‘Orchestra’ tool to decide if task needs open-source or closed-source model

Bank deferred hiring 11K call center employees thanks to Erica, exec says

Why Flagstar’s AI agents don’t have access to internet

Related Posts

FinAi Podcast

Connect

Welcome Back!

Retrieve your password

Unlock This Article

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