OpenAI has launched AgentKit, a comprehensive, no-code platform designed to streamline the creation, deployment, and evaluation of AI agents, announced at the company’s DevDay 2025 conference on October 6, 2025. Described as a “complete set of building blocks” for taking agents from prototype to production, AgentKit integrates visual Agent Builder, embeddable ChatKit UI, and enhanced Evals tools, all powered by the Responses API. For AI developers, enterprise builders, and innovators searching OpenAI AgentKit launch, no-code AI agent builder, or DevDay 2025 AgentKit, this toolkit reduces the friction of custom orchestration, prompt tuning, and frontend work, enabling rapid prototyping with drag-and-drop logic, guardrails, and versioning. CEO Sam Altman demoed building a DevDay agenda agent in under 8 minutes, showcasing its efficiency for workflows like customer support and sales agents. Available in beta for Agent Builder and generally for ChatKit and Evals, AgentKit is free during preview, with standard API pricing post-launch, positioning OpenAI against platforms like Zapier and LangChain in the agentic AI space.
AgentKit’s modular design supports everything from simple chatbots to complex multi-agent systems, with early adopters like Albertsons and HubSpot reporting 10x growth in agent deployments.
AgentKit’s Core Components: From Builder to Deployment
AgentKit consolidates fragmented tools into a unified workflow, making agent development accessible to non-coders while empowering experts.
Agent Builder: Visual No-Code Canvas
A drag-and-drop interface for composing agent logic with nodes, tools, and guardrails—preview runs, inline evals, and full versioning included.
- Key Features: Prebuilt templates, custom actions, and deployment to ChatKit or SDK.
- Demo Impact: OpenAI engineer Christina Huang built a functional agent in 8 minutes, complete with tools and widgets.
ChatKit: Embeddable UI for Agents
An interactive interface for embedding agents in apps or websites, with natural language entry points and visual consistency.
- Integration: Supports ChatGPT Enterprise/Edu and API users; admin panel for governance.
- Use Cases: Seamless chat experiences for customer service or research.
Evals for Agents: Robust Testing Suite
Expanded evaluations with datasets, automated graders, trace grading for workflows, prompt optimization, and third-party tool support.
- Metrics: End-to-end assessments, PII detection, and policy enforcement.
- Availability: Generally available, with beta features for advanced tracing.
| Component | Purpose | Availability |
|---|---|---|
| Agent Builder | Visual Workflow Design | Beta |
| ChatKit | Embeddable UI | General |
| Evals | Performance Testing | General |
Pricing and Access: Free Preview, Standard API Post-Launch
AgentKit is free during its preview phase, with Agent Builder in beta and ChatKit/Evals generally available. Post-launch, pricing aligns with OpenAI’s API models, including the Responses API underpinning the toolkit.
- Preview Access: Open to all developers; sign up via platform.openai.com.
- Post-Preview: Standard rates for Responses API (e.g., $0.002/1K tokens for GPT-4o-mini).
- Enterprise: Custom plans for large-scale deployments.
Connector Registry, for secure tool connections, begins beta rollout to API and ChatGPT Enterprise/Edu users.
Competitive Landscape: AgentKit vs. Zapier and LangChain
AgentKit differentiates with OpenAI’s native models for reasoning and tool use, contrasting Zapier’s no-code automations and LangChain’s code-first frameworks.
- vs. Zapier: Visual focus with AI-native guardrails; faster prototyping.
- vs. LangChain: Integrated evals and UI; less boilerplate code.
- Early Wins: HubSpot improved Breeze AI responses; Albertsons built an agent in hours.
Implications: Democratizing Agentic AI
AgentKit could accelerate enterprise adoption:
- Developer Efficiency: From weeks to minutes for prototypes.
- Scalability: Supports complex workflows with versioning and monitoring.
- Innovation: Enables apps in ChatGPT, blending agents with commerce.
As Altman said: “AgentKit is everything you need to build, deploy, and optimize agent workflows with way less friction.”
Conclusion: AgentKit’s Prototype-to-Production Leap
OpenAI’s AgentKit launch at DevDay 2025 is a no-code powerhouse for AI agents, blending visual building, embeddable UIs, and evals for seamless workflows. From 8-minute demos to enterprise scale, it’s a developer dream. For builders, dive in—will it redefine agent creation? The nodes connect. Tech crunch
