Every began with structured discovery interviews across the hedge fund’s teams — mapping day-to-day workflows to identify where compliance bottlenecks were occurring and, critically, to surface the inefficiency that both the compliance team and leadership had normalized. This diagnosis step was essential: neither group had recognized the review bottleneck as something solvable before the interviews surfaced it. With the problem clearly defined, Every built a custom GPT trained on the firm’s existing compliance standards and policies. Rather than adding another tool to the compliance team’s workflow, the GPT was deployed across all internal departments — giving every team the ability to self-screen content before submission. The model pre-screened materials and flagged non-compliant elements, moving submissions from approximately 30% compliant on arrival to 95% compliant. The compliance team then shifted from reviewing every piece of outgoing content to handling only the exceptions the AI flagged or could not resolve. The solution ran on existing ChatGPT infrastructure with minimal engineering overhead, moving from discovery to production in approximately four to eight weeks.