OpenAI Presence Turns Enterprise Agents Into a Managed Deployment Product
The limited-availability product combines voice and chat agents with controlled system access, simulations, escalation rules and a Codex-powered improvement loop.
OpenAI is packaging the operational work around enterprise agents—permissions, policy, evaluation and controlled updates—into a product. That matters for companies trying to move automation into real workflows, where a capable model is only one part of a system that must stay inside business rules. Announced on July 22, OpenAI Presence is available to eligible enterprise customers through limited general availability for voice and chat agents.
How Presence works
Each deployment begins with one defined job, such as resolving a billing problem, supporting an insurance claim or handling an internal IT request. The agent receives only the knowledge and system access required for that job, while the customer sets the policies for approved actions, approval points and escalation to a person.
Presence brings those controls together with standard operating procedures, guardrails, simulations and evaluation tools. Production sessions and escalations then reveal gaps. Codex, using the Presence plugin, investigates those signals and proposes updates that teams can test against the production version before approving a controlled rollout.
This makes the improvement process as important as the initial launch. OpenAI is presenting enterprise-agent deployment as a continuing operating discipline: narrow permissions, measured behavior and reviewed changes rather than an unattended model call.
Early use and the limits of the evidence
OpenAI says Presence powers its English-language phone support channel and now resolves 75% of inbound issues without human assistance. The company also says its Codex-powered improvement loop reduced human handoffs by 15 percentage points in ten days. Those are first-party operating results from OpenAI's own environment, not independent evidence that another organization will see the same outcome.
BBVA is exploring voice support for routine banking needs in Mexico, SoftBank is testing Japanese-language customer conversations, and IAG is exploring support during high-demand events such as severe weather. The announcement describes those efforts as exploring or testing, so they should not be mistaken for proof of identical production maturity across all three companies.
What changes now
Presence is a deployed product, not a self-serve dashboard. OpenAI Forward Deployed Engineers and selected global systems integrators lead implementations for eligible enterprise customers. That gives larger organizations a route to combine agents with permissions, evaluations and escalation rules, while leaving smaller teams unable to activate the product directly today.
The next meaningful checkpoint is access: whether OpenAI broadens eligibility, publishes a self-serve path or supplies evidence from customer deployments outside its own support operation.
Status
Confirmed product announcement. Availability and product structure come from OpenAI's official release; performance and partner-benefit claims remain vendor-reported.
Sources
Update note: Last reviewed 2026-07-27. We will revise this post if OpenAI changes eligibility, access or supporting evidence.
Sources
- OpenAI — Introducing OpenAI Presence — official
Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.