Chatham Financial built a trade-validation app with Codex

Chatham Financial used OpenAI’s Codex to build a trade-validation application that gathers transaction evidence, compares key terms and flags discrepancies for review, according to an OpenAI case study. The capital-markets advisory firm is comparing the application’s results with those of experienced reviewers before expanding automation.

Chatham says an early measurement found that the application cut review time from approximately 30 minutes to under four minutes. The case study did not disclose the sample size, trade mix, baseline method, error rate or acceptance criteria, or cite an independent evaluation. The same distinction applies to 1Password’s reported Codex productivity measurement: both results were reported by the companies rather than independently audited.

The application supports Chatham’s Controls and Data Integrity team, which checks whether system records match what clients authorized and what was executed. Public materials establish that Chatham used Codex to build the application, but do not say that Codex participates when the application runs trade validations.

Separately, Chatham describes Process Zero as a method that starts with the required output, pares a workflow down to its essential inputs and redesigns the process before selecting technology. In Chatham’s description of the method, subject-matter experts test redesigned processes and certify that the outputs meet the required standard.

The company says a prototype must be tested, secured, permissioned, integrated, monitored, governed, compliant and auditable before production. A named owner and subject-matter expert are responsible for the final go-or-no-go decision, based on explicit acceptance criteria.

Chatham plans to extend the trade-validation application to additional trade types and automate more of the workflow while retaining controls and professional oversight.