The Future of Enterprise AI: Empowering Engineers and Agents

The Latest Innovation in Enterprise AI Agents by OpenAI

OpenAI has introduced a new way to acquire enterprise AI agents known as OpenAI Presence. This managed product, launched on July 22, is available through a limited general availability program and is not yet accessible as a self-service offering. Deployments are overseen by OpenAI’s Forward Deployed Engineers and a select group of global systems integrators.

This marks a shift for OpenAI, which has traditionally operated on API keys and seat licenses. Presence is marketed as a project rather than a product. Each engagement begins with a specific task, such as resolving billing disputes, processing insurance claims, or managing employee IT service requests.

The AI agent is provided only with the necessary knowledge and system access for the task at hand. Customers define the rules governing the agent’s actions, sign-offs, and when human intervention is required. Post-launch, Codex analyzes production sessions and escalations, suggesting changes that the customer’s team tests and approves before implementation.

OpenAI is transparent about the extensive process involved in deploying Presence. The documentation outlines a six-stage process covering scoping business outcomes, security and privacy reviews, simulation and testing, staged rollout, and post-launch refinement. According to OpenAI, a Presence agent does not become production-ready simply by ingesting data.

The Real Problem Being Addressed

Gartner has issued a warning that over 40% of AI projects will be cancelled by the end of 2027 due to governance issues, undefined business value, and operational weaknesses rather than model capabilities. OpenAI’s Presence is designed to tackle these challenges head-on. Through simulations and grading, the agent’s performance is assessed before external interactions occur. Guardrails are in place to prevent interactions from straying beyond defined parameters. Session records and action histories provide audit trails, and controlled rollout with rollback mechanisms ensure smooth deployment.

Enterprises have learned that integrating, managing permissions, and handling changes are the most challenging aspects of deploying production agents. By offering engineering support, OpenAI addresses the actual pain points faced by customers, rather than just providing another dashboard and expecting customers to bridge the gap.

Identifying the Constraints

Access to Presence is contingent on workflow suitability, readiness for implementation, and available delivery capacity. The involvement of Forward Deployed Engineers, akin to consultants embedded in customer operations, introduces an economic aspect that differs from traditional metered inference models. While this approach works well with small volumes, scaling may pose challenges in the future.

When a vendor is also the implementation partner, clarity is essential in defining accountability for any policy misapplications in production.

The Early Stage of Enterprise AI Agents

OpenAI emphasizes that Presence has undergone rigorous testing, drawing from years of deploying agents with enterprise clients. Notable achievements include the success of OpenAI’s English-language phone support line and early collaborations with customers like BBVA, SoftBank, and IAG. These partnerships are in the nascent stages, emphasizing the ongoing development and validation of Presence.

Unveiling What Remains Unrevealed

Pricing details are not publicly disclosed, and implementation costs are tailored to each customer and deployment. The flexibility in model selection and configuration allows for adaptation as workflows evolve, although clarity on version control is crucial for evaluation purposes.

Presence operates separately from ChatGPT Workspace Agents and API access, offering customers multiple pathways to leverage OpenAI’s capabilities. The emphasis on delivery capacity as a critical factor highlights the importance of seamless integration and support.

For more insights on AI and big data, attend the upcoming AI & Big Data Expo events in Amsterdam, California, and London, co-located with other leading technology expos under TechEx. Powered by TechForge Media, these events offer valuable insights into enterprise technology trends.

Stay informed about upcoming events and webinars in the enterprise technology space by visiting here.