OpenAI Pivots to Managed Services with Presence AI Agents

Presence is not a product in the traditional sense; it is a bespoke engagement model designed to solve specific operational tasks like billing disputes or IT service requests. OpenAI’s documentation outlines a rigorous six-stage deployment process, emphasizing that these agents require extensive integration, security reviews, and custom guardrails to move into production. This strategy addresses a known industry failure point, as Gartner estimates that over 40% of agentic AI projects will be abandoned by 2027 due to poor governance and undefined business outcomes rather than technological limitations.

The Consulting Constraint

By adopting a model similar to Palantir’s Forward Deployed Engineers, OpenAI is placing significant constraints on its own growth. Unlike software, human expertise is difficult to scale, and access to Presence is currently gated by available delivery capacity and workflow readiness. While current customers like BBVA, SoftBank, and IAG serve as design partners, none are operating the agents at full scale. OpenAI cites its own internal phone support line as a success story—claiming a 75% resolution rate without human intervention—but the lack of public pricing and the reliance on custom, model-agnostic configurations leaves enterprise buyers navigating a complex, opaque procurement process.