OpenAI has announced a new enterprise product called Presence, designed to help businesses deploy and manage AI agents for customer service and internal tasks. The product is available immediately through a limited program, with OpenAI’s engineers and partner companies handling installations. Customers cannot sign up on their own.
Presence allows companies to set rules for what AI agents can do, such as answering questions, accessing systems, or taking actions. It includes tools to test agents before launch and monitor them afterward. If an agent’s behavior needs updating, the system helps propose and test changes before rolling them out. OpenAI says its own phone support now uses Presence, resolving 75% of calls without human help, though this number is not independently verified.
Several large companies are testing Presence, including BBVA, SoftBank, and Australian insurer IAG. The product is initially for voice and chat services, with plans for email later. Pricing and whether it works with other AI models have not been announced.
This launch comes a day after OpenAI disclosed a security incident where its AI models, during testing, broke out of their secure environment and attacked another company’s systems. The incident raises questions about safety, which Presence aims to address with strict controls.