OpenAI has introduced OpenAI Presence, a new managed platform designed to help enterprises deploy trusted AI agents into production workflows. Announced on July 22, 2026, Presence is now available to eligible enterprise customers through a limited general availability (GA) program, with deployments led by OpenAI’s Forward Deployed Engineers and select systems integrators (openai.com).
Presence is built to address the growing demand for reliable, high-stakes AI agents that can handle tasks such as customer support, billing resolution, insurance claims, and internal IT service requests. The platform combines model reasoning with policies, guardrails, escalation rules, and evaluation tools, ensuring agents act within defined boundaries and escalate to humans when needed (openai.com).
Each deployment begins with a specific workflow. OpenAI works closely with customers to connect relevant systems and knowledge, define permissions and policies, test agent behavior, and bring the agent into production. After launch, Presence continuously monitors performance and escalations. A Codex-powered improvement loop analyzes gaps and proposes updates, which teams can test and approve before rollout—enabling agents to adapt as business conditions evolve (openai.com).
OpenAI cites internal results from its own deployment: Presence powers its English-language phone support channel (1‑888‑GPT‑0090), where agents now resolve 75% of inbound issues without human assistance. Within 10 days, the Codex improvement loop reduced human handoffs by 15 percentage points (openai.com).
The platform is already being piloted by major enterprises: BBVA is exploring AI-powered voice support in Mexico, SoftBank is testing Japanese-language customer conversations, and IAG is evaluating support during high-demand events like severe weather (openai.com).
In summary, OpenAI Presence marks a significant step toward operationalizing AI agents in enterprise environments. By combining deployment expertise, governance, and continuous learning, it offers a practical path for organizations to scale AI agents in production with control and confidence.