The Reality of Bank GenAI
Many people assume an AI Product Manager spends all day training models or tweaking algorithms. In reality, especially in the highly regulated banking environment, my day is mostly spent bridging the gap between “bleeding-edge technology” and “conservative institutional rules.”
A typical day involves:
- 9:00 AM: Reviewing the daily token usage and latency metrics from our RAG (Retrieval-Augmented Generation) clusters.
- 10:30 AM: Meeting with the Compliance and Legal teams to explain why our LLM won’t accidentally leak PII (Personally Identifiable Information) in its chat outputs.
- 1:00 PM: Conducting a “Vibe Coding” session with the engineering team to rapidly prototype a new internal tool using AI-assisted IDEs.
- 4:00 PM: Synthesizing user feedback from Relationship Managers who tested the new proactive portfolio recommendation agent.
The True Value of an AI PM
As models become commoditized, the value of the PM isn’t in understanding the math behind attention mechanisms—it’s in seeing the friction in human workflows and knowing exactly which GenAI pattern (Classification, Extraction, Generation, Dialogue) can permanently remove it.
💬 Read more: 2025 Year in Review (English)