What Happened?

OpenAI has launched a brand-new tool called the Responses API, which helps developers create AI agents capable of performing tasks independently. The Responses API comes with built-in web search and document search features, allowing it to quickly scan relevant documents in a company’s database.

At the same time, OpenAI has released the open-source Agents SDK, which enables developers to easily orchestrate workflows among multiple AI agents and monitor and troubleshoot AI agent activities.

OpenAI Promotes AI Agent Applications with New Tool

OpenAI introduced a new tool called the “Responses API” aimed at helping businesses and developers build AI agents capable of independent task management. The Responses API module allows these agents to perform tasks like web searching, document querying, and website navigation.

Olivier Godement, the head of OpenAI’s API product, pointed out that while AI agent technology has received widespread attention in recent years, truly efficient and reliable agents are still in the early stages of development. Godement believes that demonstrating the operation of AI agents is easy, but scaling up operations and encouraging frequent usage is actually quite challenging.

The launch of the Responses API is OpenAI’s effort to offer more powerful features while advancing the application of AI agent technology towards maturity.

What are the Features of the Responses API and Agents SDK?

The Responses API includes a built-in web search tool that uses the GPT-4o and GPT-4o mini search models, allowing it to browse information from the web in real-time, accurately cite sources, and generate responses. According to OpenAI’s testing, the search accuracy of GPT-4o is 90%, while GPT-4o mini’s search accuracy is 88%.

Additionally, the Responses API includes a document search feature that can quickly scan relevant documents in a company’s database. This is particularly useful for industries such as client management or legal assistance. OpenAI emphasizes that these tools will not use the data for model training, ensuring the security of corporate data.

To assist developers in adjusting AI agents, OpenAI also released the “Agents SDK,” an open-source toolkit. This is a multi-agent collaboration framework that helps developers integrate various models during the development process, easily orchestrate workflows among multiple AI agents, and monitor and troubleshoot AI agent activities.

OpenAI stated that the Agents SDK is suitable for a variety of real-world scenarios, including customer service automation, multi-step research, content generation, code review, and sales development. For example, Coinbase uses the Agents SDK to quickly design and deploy AgentKit, enabling AI agents to interact smoothly with cryptocurrency wallets and blockchain activities.

Sources: The Verge, Tech Crunch