OpenAI agent builder launched to help developers build and ship AI agents 

At OpenAI’s Dev Day event on Monday, CEO Sam Altman announced the launch of AgentKit, a toolkit for building and deploying AI agents.

“AgentKit is a complete set of building blocks available in the open AI platform designed to help you take agents from prototype to production. Altman stated, “It is everything you need to build, deploy, and optimize agent workflows with significantly less friction.”

The launch demonstrates OpenAI’s efforts to encourage developer adoption by speeding up and simplifying agent building. It also signals a competitive move against other AI platforms racing to offer integrated tools for building autonomous agents for enterprises that can perform complex tasks, not just respond to prompts.

At OpenAI’s Dev Day, AgentKit was one of several announcements. Other announcements included the ability to build apps directly within ChatGPT, which now has 800 million weekly active users. A few fundamental features are included in AgentKit. The first is Agent Builder, which Altman compared to Canva in that it allows users to build agents. “It’s a fast, visual way to design the logic, steps, ideas,” Altman said. “It’s based on the responses API, which hundreds of thousands of developers already use.” The second capability of AgentKit is ChatKit, which provides a simple embeddable chat interface that developers can use to bring chat experiences into their own apps.

“You can bring your own brand, your own workflows, whatever makes your own product unique,” Altman said.

Evals for Agents introduces tools to measure AI agent performance, including step-by-step trace grading, datasets for assessing individual agent components, automated prompt optimization, and the ability to run evaluations on external models directly from the OpenAI platform.

Last but not least, AgentKit gives developers access to OpenAI’s connector registry, allowing them to securely connect agents to internal tools and third-party systems via an “admin control panel” while still maintaining control and security. Christina Huang, an OpenAI engineer, built a complete AI workflow and two AI agents live on stage in less than eight minutes to demonstrate how simple it is to use AgentKit. “This is all the stuff that we wished we had when we were trying to build our first agents,” Altman said, noting that OpenAI has already signed on several launch partners that have already scaled agents using AgentKit.