OpenAI CEO Sam Altman announced the launch of AgentKit, a new toolkit designed for building and deploying artificial intelligence (AI) agents, during the company's Dev Day event on Monday. The introduction aims to enhance developer adoption by simplifying and accelerating the agent development process.
Altman described AgentKit as "a complete set of building blocks available in the OpenAI platform designed to help you take agents from prototype to production." He further stated it provides "everything you need to build, deploy, and optimize agent workflows with way less friction."
The launch positions OpenAI competitively against other AI platforms that are also developing integrated tools for creating autonomous agents. These agents are intended for enterprise applications where they can execute complex tasks rather than merely responding to prompts. AgentKit was unveiled alongside several other announcements at Dev Day, including the functionality to construct applications directly within ChatGPT, which currently reports 800 million weekly active users.
AgentKit incorporates several core capabilities. The first is Agent Builder, which Altman characterized as a visual design tool for agents, akin to Canva. He noted it offers "a fast, visual way to design the logic, steps, ideas," and is constructed upon the existing responses API utilized by hundreds of thousands of developers.
Another component, ChatKit, furnishes an embeddable chat interface, allowing developers to integrate chat experiences into their own applications. "You can bring your own brand, your own workflows, whatever makes your own product unique," Altman stated regarding ChatKit's customization potential.
Evals for Agents is also included, providing tools to assess AI agent performance. This functionality features step-by-step trace grading, datasets for evaluating individual agent components, automated prompt optimization, and the capacity to conduct evaluations on external models directly from the OpenAI platform.
Finally, AgentKit offers access to OpenAI’s connector registry. This allows developers to securely link agents to internal tools and third-party systems via an "admin control panel," ensuring maintained security and control over data interactions. Christina Huang, an OpenAI engineer, demonstrated the toolkit's ease of use by constructing an AI workflow and two AI agents live on stage in under eight minutes. Altman remarked that OpenAI has already engaged several launch partners who have scaled agents using AgentKit.
