OpenAI Presence explained: OpenAI's new platform for AI customer support agents
Essential brief
OpenAI has launched Presence, a platform designed to deploy AI customer support agents that handle tasks such as billing issues, insurance claims, and IT tickets. Unlike typical chatbots, Presence
Key topics
Key facts
Highlights
Why it matters
OpenAI Presence represents a shift from static chatbot models to continuously improving AI agents that adapt based on real-world interactions. This innovation could significantly reduce the need for human intervention in customer support, improving efficiency and scalability for enterprises. The platform's controlled deployment model also highlights the importance of trust and oversight in AI applications within regulated industries.
OpenAI recently introduced Presence, a platform designed to create AI agents for enterprise customer support. Unlike traditional chatbot prototypes, Presence is built on years of enterprise deployment experience and focuses on solving specific customer service tasks such as billing inquiries, insurance claims, and IT support tickets. Each AI agent is tailored to a single task and is granted access to the necessary knowledge and systems to complete it.
Companies using Presence define policies that govern what actions the AI agent can perform autonomously, which require approval, and when to escalate issues to human agents. This ensures control and compliance with company standards. A key innovation of Presence is its continuous improvement loop powered by OpenAI's Codex coding model. When agents encounter issues or escalations during operation, Codex generates code suggestions to enhance the agent's performance, which human teams review and implement.
This method allows for ongoing adaptation to changing customer behaviors without the need for manual retraining or monthly engineering interventions. OpenAI currently uses Presence internally to manage its English-language phone support, handling tasks such as identity verification and account context analysis. According to OpenAI, the system resolves 75% of incoming support requests without human intervention and has reduced human handoffs by 15 percentage points within 10 days through Codex-driven updates. These figures are based on OpenAI's internal data and have not been independently verified.
Early enterprise customers include BBVA in Mexico, SoftBank in Japan, and IAG, who are experimenting with AI-driven voice support, natural language conversations, and high-traffic customer support scenarios. Presence is not available as a self-serve subscription product; instead, it is offered through limited general availability programs with deployments managed by OpenAI's Forward Deployed Engineers and selected systems integrators.
This approach emphasizes trust and control, incorporating pre-launch safeguards, edge case simulations, and human-in-the-loop feedback mechanisms. The effectiveness of Presence under heavy production loads, especially in regulated industries like finance and insurance, remains to be seen. OpenAI's move away from API subscriptions toward a consulting-style deployment model marks a significant shift in how AI customer support solutions are delivered.
Key topics in this update include openai presence explained openai, openai presence explained, and explained openai.