Details
- OpenAI announces Presence, an enterprise product for deploying trusted AI agents across customer-facing and internal workflows.
- Presence enables agents to answer questions, resolve issues, use company systems, take approved actions, and escalate to human staff when necessary.
- The platform combines model reasoning with enterprise policies, guardrails, and escalation rules to verify accuracy and keep agents within defined boundaries.
- Enterprises can use Presence to automate support, sales, and internal service tasks via voice and chat, including billing fixes, insurance claims, and IT requests.
- Presence includes simulations, evaluation tools, and Codex-powered improvement loops so agents can be tested against edge cases and then continually refined based on real interactions.
- OpenAI reports that Presence already powers its own English-language phone support, autonomously resolving most inbound calls and reducing human handoffs.
- The product is offered through a limited general availability program for eligible enterprise customers, delivered with OpenAI Forward Deployed Engineers and select systems integrators rather than self-serve access.
Impact
Presence extends OpenAI’s enterprise push beyond generic models into packaged agent governance, positioning it against service-focused offerings from cloud and AI rivals. By bundling guardrails, testing tools, and continuous improvement workflows, OpenAI lowers the operational barrier for large companies to put AI on the front line of customer and employee interactions while retaining tight control over risk and compliance.
