Gondola's AI gain is about prototypes, not revenue. OpenAI News says the creative hiring platform doubled revenue year over year with no added payroll after adopting Codex.
I read this as a capacity story for a very small team. Monthly engineering output rose 25% with no change in headcount, per the company, while lead engineer output rose 50% in merged changes.
That second number does the real work.
Jared Kleinstein, Gondola founder and CEO who built it after running Fresh Tape Media, says the shift is who can build. His team used Codex Cloud to explore the codebase and prototype Programs, a tool for finding creatives and tracking budgets. Creator partnerships manager Jojo Hesketh sponsored Programs and brought a working version to engineers, per the company. Kleinstein said, "A product manager doesn’t need to stop at a written product requirement anymore. They can bring engineering a working prototype."
The build started on GPT-5.4, did most work on GPT-5.5 and used GPT-6.1 for later fixes, per the company. Kleinstein estimates Codex saved days, if not weeks, and says Programs launched with fewer bugs than hand coding would have allowed. The same pattern produced a World Cup Creator Map this summer, with Hesketh building the prototype and engineers polishing it for launch while the event need was still timely.
That handoff matches how Codex is designed to work outside Gondola. OpenAI's help center notes Codex Cloud lets users start and continue coding tasks from desktop, web or mobile, separate from fast Chat assistance. An independent CLI guide describes Codex as a multi-surface agent that reads the codebase, runs commands in an OS-level sandbox, patches files and delegates tasks to the cloud. Gondola uses the CLI, editor tools and GitHub review, connected to repos and Slack, with engineers still owning architecture and security.
The fair pushback is that revenue did not double because prototypes got faster. Gondola also reports 2.5 times as many creators hired year over year, per the company, and a small team can grow fast from a low base. PR review time held steady while changes got larger, and a person still reviews consequential ChatGPT work, which kept quality from slipping.
I think the claim survives that caveat because output rose without headcount. If you are a founder asking what to copy, copy the prototype handoff, not the revenue multiple.
