OpenAI has introduced OpenAI Presence, a new enterprise product designed to help companies deploy and manage AI agents across customer service and internal business operations.
The product helps enterprises deploy AI agents that can answer questions, connect to company systems, follow internal policies, perform approved tasks, and hand complex issues to human employees when necessary, according to OpenAI.
Presence targets workflows such as customer support, insurance claims, billing problems, employee IT requests, and other business processes where companies want automation but still need control over how AI behaves.
The launch marks a shift in OpenAI’s enterprise strategy. Instead of only providing access to AI models, the company is moving into the software and operational layers required to make AI useful inside large organizations.
Built for AI agents operating in the real world
OpenAI said the biggest challenge for companies is no longer proving that AI agents can work, but making them reliable enough for production environments where rules, products, and customer expectations constantly change.
Presence combines several tools needed to operate enterprise agents, including company policies, standard operating procedures, permissions, guardrails, simulations, evaluation tools, and a Codex-powered improvement process.
“Presence brings together the components teams need to run agents in production: policies and standard operating procedures, guardrails, approved actions, simulations, evaluation tools, and a Codex-powered improvement process,” OpenAI said.
Each deployment begins with a specific business task. The AI agent receives only the information and system access needed for that role, while companies decide what actions it can take, when approval is required, and when a human should intervene.
Before launch, businesses can test agents against common requests, unusual situations, and higher-risk scenarios. After deployment, production interactions and escalations help identify problems, while Codex can suggest improvements that employees review before changes go live.
Already powering OpenAI’s own support operations
OpenAI said Presence is already being used for its English-language phone support service, where the system handles customer conversations, verifies callers, accesses account information, and performs approved actions.
According to the company, the system now resolves 75% of inbound issues without human assistance. OpenAI also said its Codex-powered improvement process reduced human handoffs by 15 percentage points over 10 days. Several companies are also exploring the platform.
BBVA is testing AI-powered voice support for banking services in Mexico, SoftBank is testing Japanese-language customer conversations, and IAG is exploring support use cases during high-demand events such as severe weather.
The road ahead for enterprise AI agents
The biggest test for products like Presence will not be whether AI agents can complete individual tasks, but whether they can operate safely and consistently inside complex businesses.
Enterprises often need strict controls around customer data, compliance requirements and decision-making authority. OpenAI’s focus on simulations, evaluations and approval-based updates shows the company understands that businesses are unlikely to hand over critical workflows to fully autonomous systems.
However, the high-touch deployment model could limit adoption among smaller companies that want faster, self-service tools. The lack of public pricing and broader availability details also leaves questions about how accessible Presence will become.
Also read: Google launched Gemini 3.6 Flash and Gemini 3.5 Flash-Lite for lower-cost enterprise workloads, while its cybersecurity model remains restricted from public use.