OpenAI today (July 22, 2026) introduced OpenAI Presence, a new enterprise-grade product designed to help organizations deploy AI agents into production with a focus on trust, control, and adaptability. Presence is available immediately for voice and chat agents and is already proven in high-stakes environments such as customer support and internal workflows.(openai.com)
Presence combines model reasoning with enterprise-grade governance. Each deployment is scoped to a specific job—such as billing inquiries, insurance claims, or IT support—and includes only the necessary knowledge and system access. Companies define policies for agent behavior, escalation triggers, and approval workflows. After launch, the system uses production data and escalations to identify gaps, with Codex-powered suggestions for improvements that teams can test and approve before rollout.(openai.com)
OpenAI cites real-world performance: its own English-language phone support channel, powered by Presence, now resolves 75% of inbound issues without human assistance. Within 10 days of deployment, human handoffs dropped by 15 percentage points thanks to the Codex-driven improvement loop.(openai.com)
Presence is available now to eligible enterprise customers through a limited general availability program. Deployments are supported by OpenAI’s Forward Deployed Engineers and select systems integrators. The platform is built to evolve with enterprise needs, incorporating insights from each deployment into ongoing research and product development.(openai.com)