OpenAI releases Presence for eligible enterprise customers

• OpenAI Presence is a limited-GA enterprise product for voice and chat agents that can answer questions, use company systems, take approved actions, and escalate to people, according to OpenAI's launch post.
• Deployments start from a specific job, with only the knowledge and system access needed for that job, as TheRealAdamG's excerpt notes.
• The product bundles the production harness around the model: policies, guardrails, escalation rules, simulations, evals, approved actions, and a Codex-powered improvement loop, per TheRealAdamG's excerpt.
• OpenAI says its own English-language phone support channel now resolves 75% of inbound issues without human assistance, a proof point described in OpenAI's launch post.
• Presence is not self-serve; btibor91's screenshot says deployments are led by OpenAI Forward Deployed Engineers and select global systems integrators.

OpenAI's official post includes the spicy production numbers: 75% autonomous issue resolution on its English-language phone support line, plus a 15 percentage point drop in human handoffs within 10 days. The same page names BBVA, SoftBank, and IAG as early enterprise users. The availability page in btibor91's screenshot is the practical caveat: Presence is limited GA, eligible enterprise only, and not available as a self-serve product.

Presence deployments begin with a named job, not a general workplace agent. OpenAI's official post gives three starter shapes:

• Resolving billing issues.
• Supporting insurance claims.
• Resolving employee IT service requests.

The access model is least-privilege by design: the agent receives only the knowledge and system access required for that job. In a billing workflow, OpenAI says the agent can understand the request, verify the customer, look up account information, apply company policy, and take an approved action.

Presence is OpenAI's attempt to productize the boring parts of production agents. The stack OpenAI lists is the stuff engineers usually have to assemble around the model:

• Policies and standard operating procedures.
• Guardrails.
• Approved actions.
• Escalation rules.
• Simulations.
• Evaluation tools.
• A Codex-powered improvement process.

Before launch, simulations and graders test whether the agent reached the right outcome, followed policy, used tools correctly, and escalated when appropriate, according to OpenAI's trust section.

After launch, production sessions and escalations become training data for the deployment process. OpenAI says Codex, using the Presence plugin, investigates quality signals and suggests updates.

Teams can test each proposed change against the production version before approving a controlled rollout. That is the most concrete implementation detail in the launch, because it turns agent maintenance into a review-and-approve loop rather than a silent prompt drift problem.

OpenAI says Presence powers its English-language phone support channel at 1-888-GPT-0090. The agent handles open-ended requests, verifies callers, uses account context, and takes approved actions.

The reported launch metrics are first-party, but useful:

• Frontline support quality: met or exceeded OpenAI's internal benchmarks within weeks.
• Inbound issue resolution: 75% resolved without human assistance.
• Human handoffs: reduced by 15 percentage points in 10 days through the Codex-powered improvement loop.

OpenAI's customer section names three early enterprise tracks:

• BBVA is exploring AI-powered voice support for everyday banking needs in Mexico.
• SoftBank is testing natural Japanese-language customer conversations.
• IAG is exploring support during high-demand events such as severe weather.

Daniel Ordaz, Head of AI Transformation at BBVA Mexico, said BBVA is working with OpenAI as a Presence design partner for financial customer-service voice experiences.

Presence ships through limited general availability for eligible enterprise customers. AILeaksAndNews' launch-day post said the product would be available for voice and chat agents later that day, while OpenAI Devs routed developers to the same product page in its launch link.

The access model is services-heavy: OpenAI Forward Deployed Engineers and select global systems integrators lead deployments. A separate AILeaksAndNews thread also preserved the voice-and-chat launch wording in its Presence thread context.

OpenAI says it will keep supporting voice customers with access to frontier models through the OpenAI API. VentureBeat's launch coverage reported that OpenAI had not disclosed pricing, contractual terms, or the cost of the engineering and integration work around Presence.