Goldman Engineers Teach AI Agents Firm's Tribal Knowledge

Goldman Sachs' engineers are now tasked with teaching AI agents the firm's unique 'tribal knowledge'. This is in addition to making their tools specific to Goldman. The bank has over 12,000 developers using AI, and they must figure out how to mentor these agents as they would a new employee.

The transfer of institutional knowledge into the AI is the biggest challenge, according to Marco Argenti, chief information officer. He said that an experienced GS AI looks different from a naive AI. The firm has drafted 'skills', reusable bundles of instructions for performing specific tasks, to capture developers' technical knowledge about Goldman.

These skills teach AI what good cloud migration looks like, specifically at Goldman. Thanks to the skill, AI tools are more helpful to those tasked with cloud migration. Argenti said that the unwritten rules are the ones that are actually harder to capture, which is why they try to systematize them.

Goldman has studied its internal processes by conducting interviews and analyzing outputs. The bank regularly updates its skills as its internal processes evolve, creating a constant loop of improvement.