Introducing OpenAI Presence

OpenAI announced OpenAI Presence, a battle-tested product for deploying trusted AI agents across customer and internal workflows. The product helps enterprises deploy AI agents that can answer questions, resolve issues, use company systems, take approved actions, and escalate to people when needed. OpenAI Presence was developed in collaboration with OpenAI's Research team and is available today for voice and chat agents. It supports real-time experiences such as customer support, outbound sales, and high-risk internal workflows. Presence pairs model reasoning with policies, guardrails, and escalation rules to verify accuracy and performance. Each deployment starts with a specific job, such as resolving billing issues or supporting insurance claims. The company sets policies on what the agent can do, when approval is needed, and when a person should take over. After launch, production sessions and escalations reveal gaps, and Codex proposes updates that teams can test and approve. OpenAI works alongside customers to identify high-value workflows, connect knowledge and systems, establish permissions, test the agent, and bring it into production. As deployment expands, OpenAI and select systems integrators can continue to support it. Presence was built through years of deploying agents with customers in mission-critical environments. It powers OpenAI's English-language phone support channel at 1-888-GPT-0090, handling open-ended requests and resolving 75% of inbound issues without human assistance. Within weeks, its Codex-powered improvement loop reduced human handoffs by 15 percentage points in just 10 days. Leading enterprises including BBVA, SoftBank, and IAG are exploring or building on Presence for various use cases. BBVA is exploring AI-powered voice support for everyday banking needs in Mexico. SoftBank is testing natural Japanese-language customer conversations. IAG is exploring timely support during high-demand events such as severe weather. Before deployment, teams can test Presence against common requests, edge cases, and higher-risk scenarios using simulations and graders. Guardrails can intervene when interactions move outside company boundaries. Production sessions and quality signals help identify where the agent needs attention as policies, products, and user behavior change. Codex investigates those signals and suggests updates, which teams can test and approve for controlled rollout. Presence is designed to learn with the business while maintaining control, improving as models and capabilities evolve.