OpenAI has introduced "Presence," a new product designed to help enterprises deploy reliable AI agents capable of handling high-value production workflows such as customer support and internal IT requests. The system combines model reasoning with specific policies, guardrails, and escalation rules to ensure accuracy and control.
- Presence supports real-time voice and chat experiences, allowing agents to resolve issues like billing or insurance claims while adhering to company-defined permissions.
- A Codex-powered improvement loop analyzes production sessions and escalations to suggest updates, enabling agents to adapt as user behavior changes.
- The product is currently available to eligible enterprise customers through a limited general availability program led by OpenAI Forward Deployed Engineers.
The announcement highlights that the underlying technology already powers OpenAI's English-language phone support channel, where it resolves 75% of inbound issues without human assistance.