OpenAI has announced the release of new tools and APIs that simplify the development of AI agents for developers and businesses. These agents can autonomously perform tasks on behalf of users, utilizing advanced logic, multimodal capabilities, and enhanced security measures. The updates include the Responses API, built-in tools such as web search and file search, as well as the open Agents SDK.
🧑💻 Computer use—powered by the same CUA model behind Operator, you can now build agents capable of completing tasks on a computer, such as web QA testing or data entry.
— OpenAI Developers (@OpenAIDevs) March 11, 2025
Available to select developers on usage tiers 3–5 as a research preview: https://t.co/nozhG5RLVg pic.twitter.com/ef2UxA5ol7
The Responses API serves as a universal foundation for agent applications. It combines the simplicity of the Chat Completions API with the tool-using capabilities of the Assistants API. Developers can leverage built-in tools such as web search, file search, and computer use to enable agents to interact with real-world data. The API supports streamlined features like intuitive streaming events and SDK helpers for accessing model outputs.
The built-in tools include: web search, which enables real-time information retrieval with embedded links—useful for applications such as research agents or shopping assistants. It runs on GPT-4o models and achieves high accuracy rates. File search facilitates queries across large document databases with metadata filtering capabilities. Computer use allows for automating browser workflows by capturing mouse and keyboard actions.
The Agents SDK streamlines the orchestration of multi-agent processes, providing features such as configurable LLMs, intelligent agent handoffs, guardrails for safety checks, and tracing tools for debugging workflows. The SDK is already being successfully used in real-world scenarios by companies like Coinbase and Box to enhance customer support automation and data generation.
