OpenAI Introduces Presence

OpenAI Presence is an enterprise product designed to let companies deploy AI agents into real-world use for customer interactions and internal workflows. Going beyond simply providing a model, it integrates policy configuration, guardrails, escalation rules, evaluation tools, and a Codex-based automatic improvement loop.

Each deployment begins as a specific task unit, such as resolving billing issues, assisting with insurance claims, or handling IT service requests. Agents are granted only the knowledge and system access needed for that task, and companies directly define the scope of the agent's actions and when human intervention should occur. After deployment, Codex proposes improvements based on production session data, and teams approve them, enabling continuous optimization.

It currently supports voice and chat channels, and can be applied to customer support, outbound sales, and internal high-risk workflows. When applied to OpenAI's own English phone support (1-888-GPT-0090), it resolved 75% of inbound inquiries without human intervention, and reduced human handoffs by 15 percentage points within 10 days.

Initial partners include BBVA (voice banking in Mexico), SoftBank (Japanese-language customer support), and IAG (support for high-demand situations such as weather events).