OpenAI has officially launched Presence, a new platform designed for deploying AI agents in corporate environments. The solution focuses on automating customer service and internal business processes, offering companies a tool for creating autonomous digital assistants.

Unlike traditional chatbots, Presence uses advanced language models integrated with access control and security systems. The agents can work with internal databases, perform authorized actions (such as refunds or insurance claims processing), and transfer the conversation to a human operator when necessary.

Key Platform Capabilities

  • Access Control: The agent receives a strictly limited set of data and permissions required to perform a specific task. This reduces the risk of information leaks.
  • Automatic Improvement: The built-in Codex system analyzes gaps in dialogues and suggests algorithm updates to improve response accuracy.
  • Multimodality: Support for real-time text chats and voice channels, expanding use cases.

OpenAI has already implemented Presence in its own English-language telephone support service. The results are impressive: the AI independently handles 75% of incoming requests. The use of Codex reduced the frequency of call transfers to humans by 15% in just 10 days. This demonstrates that the system not only solves basic issues but also quickly adapts to complex scenarios.

Among the first corporate clients are BBVA bank, telecommunications giant SoftBank, and insurance group IAG. These companies are already using Presence to automate banking transactions and handle peak loads in support services. This collaboration confirms that the technology is in demand in high-load sectors.

At the initial stage, the product is available to a limited number of corporate customers. Implementation is carried out by OpenAI engineers and certified partners. This is a strategically sound move: quality control at the start is critical for reputation in the B2B market.

My view: Presence is not just another chatbot but a full-fledged platform for building AI agents with elements of role-based control and self-learning. The 75% rate of independent query resolution and a 15% reduction in operator load within 10 days is a strong argument for corporate clients, especially amid staff shortages. If OpenAI can scale this model without losing quality, we will witness a shift toward fully autonomous corporate communications. However, data security and integration with legacy systems remain key challenges.