The Key Facts
OpenAI frames Presence around a specific pain point: enterprises no longer need to prove agents *can* work—they need agents reliable enough for high-value production work. It combines model reasoning with policies, guardrails, and escalation rules, where the company sets the boundaries for approved actions, decides when a human must approve, and when a case is handed off. Each deployment starts from one concrete job (billing issues, insurance claims, employee IT tickets) and gets only the information and system access that task requires. Before launch, teams run simulations and graders against routine requests, edge cases, and higher-risk scenarios—checking for the right outcome, correct tool use, policy compliance, and appropriate escalation. In production, a Codex-powered improvement loop analyzes live data, investigates issues, and recommends changes, but every change requires a human signature before it reaches production.
Background and Track Record
OpenAI dogfooded it first: Presence now runs its English-language phone support, verifying callers, using account context, and taking approved actions. The company says it resolves 75% of inbound calls without a human, and the improvement loop cut human handoffs by 15 percentage points in 10 days. Early pilots include BBVA (Spanish-language banking support in Mexico) and SoftBank (Japanese-language deployments).
Why It Matters
This is another step in OpenAI's shift from "selling models" to "selling managed enterprise software" (following the OpenAI Deployment Company set up in May). But Presence is not self-serve: deployments are led by OpenAI's Forward Deployed Engineers and select systems integrators, and pricing, geography, and contract terms are undisclosed. For buyers it means a more complete, governable system—but also a heavier, pricier, more vendor-dependent one worth evaluating with commercial terms surfaced early.
via: OpenAI official announcement "Introducing OpenAI Presence" (2026-07-22), with reporting by VentureBeat; verified 2026-07-25