OpenAI Launches Presence, an Enterprise Agent Platform Deployed by Consultants, Not APIs

OpenAI has launched Presence, an enterprise platform for deploying AI voice and chat agents into mission-critical workflows — and in a sharp break from its API-first history, the company is selling it the way Accenture sells transformation projects: through embedded engineers, with no self-service tier.

Agents with guardrails, sold with humans attached

Presence lets organizations build agents that answer customer questions, resolve technical issues, operate proprietary software, execute approved actions and escalate to humans — all wrapped in organizational policies, operational guardrails and audit trails. A Codex-powered continuous improvement loop tunes agents against real interactions. Target workloads span customer support, outbound sales development, procurement, IT services and HR.

The proof point OpenAI leads with is its own: Presence runs the company's English-language phone support line, where it resolves 75 percent of inbound calls without human intervention. Early adopters include Spanish bank BBVA, airline group IAG — and SoftBank, extending the Japanese conglomerate's sprawling entanglement with OpenAI from capital into day-to-day operations.

The consulting turn

Presence is available only through a limited general-availability program, with deployments led by OpenAI's Forward Deployed Engineers and select global systems integrators. The approach mirrors Microsoft's "Frontier Company" playbook — which has put some 6,000 embedded engineers inside customers — and reflects a hard-won industry lesson: an estimated 95 percent of enterprise AI pilots fail on integration, not model capability.

It also positions OpenAI in an increasingly crowded field. Google is pushing Gemini Enterprise, Meta has its Business Agent, and Nvidia and ServiceNow are building Project Arc. The battleground has shifted from model quality to governance, distribution and switching costs — and boots-on-the-ground deployment creates far stickier customers than an API endpoint ever did.

The launch, days after OpenAI announced its $20 billion Project Camellia data center campus in Georgia, sketches the company's next act: not just selling intelligence by the token, but owning the enterprise deployment layer where the margins — and the lock-in — live.