OpenAI and Nvidia turn AI agents into co-workers

Wiring the app into per‑employee cloud VMs with a zero‑data‑retention posture and read‑only access to production systems, turned AI agents turned into safe, auditable coworkers rather than experimental sidecars, says Nvidia.

The result: debugging cycles shrinking from days to hours, multi‑file refactors and complex experiments happening overnight, and entire features shipping from natural‑language prompts — all on infrastructure that delivers dramatically lower cost per token and higher throughput at rack scale.

‘This story is a window into what “real” enterprise AI looks like when a frontier model vendor and a silicon leader co‑design the full stack, from GPUs to agents,’ says Nvidia.

OpenAI’s feedback influences Nvidia’s hardware roadmap which resulted in the first GB200 NVL72 100,000-GPU cluster. 

See all our AI content.