OpenAI is betting that enterprise customers need more than just powerful models. They need someone to deploy them.
The company launched Presence on Wednesday, a platform that handles the messy work of putting AI agents into production. Rather than selling API access and leaving integration to customers, OpenAI deploys its own forward-deployed engineers alongside global systems integrators to build voice and chat agents inside existing business workflows.
Each agent gets a narrowly defined job: resolve a billing dispute, handle an insurance claim, or field IT support tickets. Company policies determine what the agent can do independently, which actions require human approval, and when a person must take over. Before going live, teams run simulations against policy changes and edge cases to catch failures before customers see them.
The product arrives as enterprises increasingly struggle to move AI agents from demo to production. Research shows a majority of companies have shipped agents that passed internal testing only to fail in real customer interactions.
Pricing remains undisclosed, and OpenAI has not confirmed whether Presence supports models from other providers. The product is available now through a limited general availability program, with broader rollout expected later this year.
