Nom
Give AI agents real UI powers without giving up control.
Start new thread

Nom - Give AI agents real UI powers without giving up control.

by•
Nom connects AI tool calls to the React components already in your app. Define what an agent can do with typed, validated tools, map the results into component props, and keep your application in control of data fetching, authorization, and rendering. It’s model-provider neutral, works with AI SDK 6, and supports loading, empty, error, and stale-request states out of the box. Build AI-powered dashboards, product search, and interactive interfaces using the components your users already know.

Add a comment

Replies

Best
Maker
📌
Hey Product Hunt! 👋 I built nom because AI features often force a choice: let the model generate UI you don’t fully control, or hand-build an integration for every possible action. Nom takes a different route. Your React components stay yours, along with your data fetching, authorization, and rendering. You simply expose the actions an agent can take as typed, validated tools, then map the results directly into your existing component props. It’s designed for teams building AI-powered products that still need predictable, app-owned UI. I’d love to hear what you’re building with AI in React, and where the integration feels most painful today. Happy to answer anything!

One thing I'd love is a built-in way to test tool definitions with mock responses right in the browser. Something like a playground where you can swap providers, fire a tool call, and see the mapped props update live would save so much back-and-forth during development.

Typed tools are a nice touch, but it would be great to see a visual debugging panel that logs each tool call and its mapped props in real time. Right now I imagine figuring out why a component received stale or empty data means digging through logs manually, especially across nested components.