At OpenAI's Developer Day event on Monday, CEO Sam Altman announced the release of AgentKit, a toolkit for building and deploying artificial intelligence (AI) agents. Altman said, "AgentKit is a complete set of building blocks available on the OpenAI platform, designed to help you take your agents from prototype to production. It includes everything you need to build, deploy, and optimize your agent workflow, significantly reducing friction."
The launch of AgentKit underscores OpenAI's efforts to increase developer adoption by simplifying the agent-building process and improving development efficiency. It also marks a significant step in OpenAI's competitive landscape: Other AI platforms are vying to offer integrated tools to help companies build autonomous agents that can perform complex tasks, rather than just responding to prompts.
AgentKit was one of several announcements at OpenAI's Developer Day, including support for developing applications directly within ChatGPT, which currently boasts 800 million weekly active users.
AgentKit includes several core features, including:
Agent Builder: Altman likens it to "Canva for agents." He explains, "It's a fast, visual way to design the logic, steps, and core ideas of an agent, built on a responsive API already used by hundreds of thousands of developers."
Chat Toolkit: Provides a simple, embeddable chat interface that developers can use to integrate the chat experience into their own apps. Altman says, "Developers can incorporate their own branding and customize workflows to create chat features with unique product attributes."
Agent Evaluation Tool: Provides tools for measuring the performance of AI agents, including step-by-step scoring, datasets for evaluating individual agent components, automated optimization prompts, and the ability to evaluate external models directly on the OpenAI platform.
Connector Registry Access: AgentKit gives developers access to OpenAI's connector registry, enabling them to securely connect their agents to internal tools and third-party systems through the Admin Dashboard, ensuring data security and user control.
To demonstrate AgentKit's ease of use, OpenAI engineer Christina Huang demonstrated how to build an entire AI workflow and develop two agents in under eight minutes.
