OpenAI says Proaction's fleet agents boost sales 60%
Proaction built voice and computer-use agents on Codex, GPT-Live-1 and GPT-6 Astra to call drivers and repair shops, OpenAI said in a customer case study.
OpenAI published a case study saying Proaction, a fleet-management startup, built agents on Codex, GPT-Live-1 and GPT-6 Astra. The company said the agents raised Proaction's sales 60% and saved more than 75 hours of staff time, figures that are OpenAI's own.
GPT-Live-1, OpenAI's voice model, lets the agents call drivers and repair shops and help arrange vehicle maintenance. A person steps in when needed, OpenAI's developer account said separately on X. GPT-6 Astra, the company's computer-use model, is credited with shortening the automated runs needed to finish the same tasks.
https://x.com/OpenAIDevs/status/2103528579212669374
OpenAI did not say how it measured the sales and hours figures, over what period, or how large Proaction is. Case studies like this one are marketing material commissioned by OpenAI, and the results may not be typical of other customers.
GPT-6 Astra is part of OpenAI's GPT-6 line of models built for agentic, computer-using tasks, distinct from GPT-Live-1's voice-focused design. Proaction is the latest customer OpenAI has featured pairing the two models with Codex, its coding tool, in one workflow.