Chatham Financial has cut trade validation time from 30 minutes to under 4 minutes by deploying OpenAI's Codex and GPT-4.5 across its capital markets workflows. The firm, which advises clients on interest rate and currency risk, is using these models to automate processes that previously required senior analyst intervention.
The more interesting story is not the time savings but the workflow redesign underneath it. Chatham is not layering AI onto existing processes. It is rebuilding how capital markets expertise gets delivered, using Codex to generate and review code that handles complex financial instrument validation. That distinction matters if you work in any domain where institutional knowledge has historically been the moat.
Read the full piece for the specifics on how Chatham structured the human-in-the-loop checkpoints and where the models still fall short. The failure cases are more instructive than the wins.
[READ ORIGINAL →]