On Monday, October 6, 2025, OpenAI launched AgentKit, a toolkit that allows developers to easily build and deploy AI agents. Until now, building agents required multiple tools and complex processes. AgentKit aims to make the entire process simpler, faster, and more interactive.
2. New Announcements at the Dev Day Event
AgentKit was announced at OpenAI’s Dev Day 2025, introducing new features such as accessing apps within ChatGPT. This means developers can now directly connect their AI agents to different apps and services within ChatGPT.
3. Previous Challenges and Complexities
Previously, developers had to deal with multiple tools and manual processes to build AI agents. This involved creating custom connectors, manual evaluation, prompt tuning, and weeks of frontend development. This complexity made creating agents difficult and time-consuming for new developers.
4. Visual Workflow Design
With AgentKit, developers can now design workflows visually. This means that agent logic and interactions can be easily created with a drag-and-drop approach, enabling faster embedding of agent UIs and reduced development time.
5. Agent Builder: The main tool for creating agents
The Agent Builder tool allows developers to create new AI agents. Users can easily set agent behavior, customize prompts, and fine-tune how agents function. Currently available in beta, even beginning developers can easily experiment with it.
6. Connector Registry: Managing Data and Tools
Connector Registry allows admins and developers to easily manage their data and tools across OpenAI products and ChatKit. This feature is currently rolling out in beta to select API users and ChatGPT Enterprise/Edu users.
7. ChatKit: Embedding Custom Chat Experiences Made Easy
ChatKit’s feature allows developers to embed custom agentic chat experiences directly into apps. This gives users the opportunity to add personalized AI chat interactions to their applications, making agents more interactive and useful.
8. Evals for Agents: New Performance and Improvement Tools
AgentKit now includes Evals for agent features, such as datasets, trace grading, automated prompt optimization, and third-party model support. This is intended to measure and improve the performance of AI agents. This helps developers identify where their agents are performing well and where improvements are needed.
9. Improving AI Agent Performance
Evals features allow developers to perform data-driven analysis of their AI agent performance. This allows them to improve agent response times, accuracy, and the quality of user interactions. This allows AI agents to continuously learn and improve.
10. Developer-Friendly and Time-Saving Toolkit
Overall, AgentKit makes AI agent development much more accessible and faster. Whether you’re a new developer or an experienced one, this toolkit simplifies agent design, integration, and performance improvement. This makes working on AI projects more fun and less time-consuming.
