OpenAI launches Presence, an enterprise platform for voice and chat AI agents

OpenAI launched Presence on Wednesday, an enterprise platform for deploying AI voice and chat agents to handle customer-support and internal-service work. The company said the system already resolves about 75% of OpenAI’s own inbound support calls without a human.

The pitch is as much about governance as capability. Presence bundles the controls enterprises tend to demand before letting an autonomous agent talk to customers: written policies and standard operating procedures, guardrails, approved-action limits, pre-launch simulations and evaluation tools. A Codex-powered improvement loop reviews production conversations and suggests behavioral changes, which a human approves before they take effect.

OpenAI said that feedback loop cut human handoffs by 15 percentage points within 10 days of a deployment, according to its announcement. The 75% resolution figure comes from OpenAI’s own English-language phone support line.

Those numbers are the company’s own and have not been independently verified; they describe OpenAI running its own tooling on its own support desk, which is a friendlier test than a third party’s messy call center. Agent platforms across the industry routinely report high containment rates in vendor demos that soften once real customers and edge cases arrive.

Presence ships through a limited general-availability program for enterprise customers only, not as a self-service product. Deployments are led by OpenAI Forward Deployed Engineers and select systems integrators, a high-touch model that signals OpenAI expects heavy customization per client.

The launch puts OpenAI in more direct competition with the contact-center and agent vendors already selling AI support tools. Whether Presence’s containment rates hold up outside OpenAI’s own phone line, and across languages beyond English, is what enterprise buyers will test first.