OpenAI has unveiled OpenAI Presence, a new enterprise-grade platform that enables organizations to deploy AI agents for complex, mission-critical operational tasks. The platform allows companies to build intelligent systems capable of answering customer inquiries, resolving technical issues, interacting with proprietary software, executing approved actions, and escalating to human operators when necessary. Presence combines advanced model reasoning with strict organizational policies, operational guardrails, and clear escalation protocols to ensure accuracy and accountability at scale. The launch represents OpenAI's most direct push into full-stack enterprise operations — moving beyond API access and ChatGPT Enterprise subscriptions toward managed, production-grade agent deployments that compete with offerings from Salesforce, Google, and Microsoft in the rapidly forming market for autonomous customer-facing AI agents.

Each deployment starts with a narrowly scoped role — such as processing billing inquiries, supporting insurance claims, or managing internal IT service requests. Agents are granted access only to the knowledge bases and enterprise systems required for that specific workflow. Company administrators retain full control over operational boundaries, approval thresholds, and handoff criteria. After launch, production interactions and escalations reveal performance gaps, which OpenAI's Codex system analyzes to propose targeted updates. Engineering teams can then test and approve these refinements, enabling the agent to adapt as customer behavior and business needs evolve. This continuous learning loop directly addresses a longstanding limitation of traditional customer service automation, where rule-based chatbots and IVR systems have historically struggled to improve after deployment and routinely fail on complex or ambiguous requests.

OpenAI works directly with enterprise clients to map high-value workflows, integrate necessary systems, establish permissions, and conduct rigorous testing before agents go into live production. This white-glove deployment model — staffed by Forward Deployed Engineers — mirrors the approach used by enterprise software companies such as Palantir, reflecting the high stakes of placing autonomous AI in regulated, customer-facing environments. The company notes that insights from every deployment inform ongoing research and product development, producing compounding benefits across its customer base.

Real-World Validation and Controlled Availability

Presence has already proven its effectiveness in live environments. Most notably, it powers OpenAI's own English-language phone support channel at 1-888-GPT-0090, where the system handles open-ended caller requests, verifies identities, accesses account context, and executes approved actions. Within weeks of deployment, the platform met or exceeded internal benchmarks for frontline human support quality and now autonomously resolves approximately 75 percent of inbound issues. Through its Codex-powered improvement loop, Presence reduced human handoffs by 15 percentage points in just 10 days. That resolution rate stands well above levels typically achieved by legacy automated support systems, which often route a majority of inquiries to human agents.

Several major global enterprises are already building on the platform. Spanish financial institution BBVA is exploring AI-powered voice support for everyday banking needs in Mexico. Telecommunications giant SoftBank is testing natural Japanese-language customer conversations. Insurance group IAG is evaluating the system for delivering timely assistance during high-demand events such as severe weather — a use case where call volume spikes and staffing constraints have historically degraded service quality during the moments customers need it most.

Before reaching end users, Presence deployments undergo rigorous simulation against common requests, edge cases, and higher-risk scenarios. Automated graders evaluate outcomes, policy adherence, and escalation appropriateness. Post-launch, production signals continuously inform refinements suggested by Codex, which engineering teams can test against live versions and approve through controlled rollouts.

OpenAI Presence is currently available to eligible enterprise customers exclusively through a limited general availability program. Deployments are led by OpenAI Forward Deployed Engineers and select global systems integrators, who work alongside clients to identify workflows and manage production rollout. The company confirmed that the product is not yet offered as a self-serve solution, though organizations can inquire through their existing OpenAI account teams. The controlled rollout suggests OpenAI is prioritizing reliability and reputation over rapid scaling — a notable contrast to the broad, self-serve availability of its consumer products.

Source: Metaverse Post