OpenAI has announced Presence, a product designed to help enterprises deploy AI agents in production while maintaining control, accuracy and adaptability. According to OpenAI, the challenge for businesses is no longer proving that agents can work but ensuring they are reliable enough to perform high-value tasks as products, policies and user behavior evolve.
What Presence does
Presence is a battle-tested solution that pairs model reasoning with policies, guardrails and escalation rules to verify accuracy and performance. Deployments begin with a specific job such as resolving billing issues, supporting insurance claims, or handling employee IT service requests. Each agent receives only the knowledge and system access required for that task, and the company defines the policies: what the agent may do, when it needs approval, and when a person should take over.
Operation and continuous improvement
After launch, production sessions and escalations reveal gaps. Codex proposes updates based on those signals; teams can test and approve changes so the agent adapts as customer behavior shifts without losing enterprise control.
OpenAI works alongside each customer to identify high-value workflows, connect required knowledge and systems, establish permissions and policies, test the agent, and bring it into production. As deployments expand, OpenAI and select systems integrators can continue support.
Concrete results and use cases
Presence supports real-time experiences across voice and chat channels — for example, customer support, outbound sales, and high-risk internal workflows. OpenAI uses Presence in its English-language phone support channel at 1-888-GPT-0090: the agent handles open-ended requests, verifies callers, uses account context, and takes approved actions.
Within weeks of launch, that channel met or exceeded the benchmarks OpenAI uses to grade frontline human support quality and now resolves 75% of inbound issues without human assistance. OpenAI reports that the Codex-powered improvement loop reduced human handoffs by 15 percentage points in just 10 days during rollout.
Safety, testing and ongoing oversight
Before a Presence deployment reaches users, teams can test it against common requests, edge cases and higher-risk scenarios. Simulations and graders check whether the agent reached the right outcome, followed policy, used tools correctly, and escalated when appropriate. Guardrails can intervene if an interaction moves outside the company’s boundaries.
Work continues after launch: production sessions, escalations and quality signals show where an agent succeeds and where it needs attention as policies, products and user behavior change. The Codex Presence plugin inspects these signals and suggests updates. Teams can test each proposed change against the production version, then approve a controlled rollout.
Availability and support
OpenAI Presence is available to eligible enterprise customers through a limited general availability program. Deployments are led by OpenAI Forward Deployed Engineers (FDEs) and select global systems integrators. Presence is not yet available as a self-serve product.
OpenAI will continue to support voice customers with access to its frontier models via the OpenAI API. Organizations interested in exploring Presence are advised to contact their OpenAI account team.
Summary
Presence aims to give enterprises a unified foundation — models, policies, guardrails and an improvement loop — for running AI agents in production with ongoing oversight. The product builds on years of real-world deployments, and insights from those projects feed back into OpenAI’s research and product development to benefit customers over time.
