Interview, Fireside Chat
BofA's Hari Gopalkrishnan on AI Strategy with Julie Hyman
- Bank of America expects to scale AI-enabled client meeting management tools from 25,000 to 50,000 teammates, following a 500-person proof of concept in December.
- A credit underwriting pilot solution is targeted for deployment at scale by the end of the current year.
- Relationship managers utilizing new tools are projected to increase client coverage from X to X plus four based on established leverage models.
- The firm plans to hire 4,000 individuals for the upcoming campus class, maintaining total hiring levels similar to the 20,000 hires made this year while managing attrition opportunistically.
- Hiring is not expected to decelerate in the foreseeable future due to significant ongoing work regarding new AI-related threats and evolving client requirements.
- Deployment will utilize a hybrid infrastructure strategy combining on-premises GPUs, cloud-based infrastructure, open-source models, and frontier models tailored to specific use cases.
- The firm intends to employ a stepwise evaluation process for tools to ensure return on investment and mitigate risks associated with rising costs like token expenses.
- Development of "Erica 2.0" aims to significantly enhance customer and user friendliness compared to the first generation of the digital assistant.
- Generative capabilities will continue to be integrated internally with a human-in-the-loop risk management framework, leveraging internal feedback to refine future external deployments.
- The strategy relies on a virtuous cycle of learning from associates through listening sessions, iterating on designs, and executing large-scale client rollouts.