The open-source Agentic UI SDK: your buttons, links and forms become tools an AI agent can call. The agent lives inside your app, so users just ask, in text or by voice, with nothing to install on their side.
The page exposes what it means and what it can do. The model reasons about those intentions and acts through them, never by taking control of the DOM.
Components register their tools when they appear and unregister them when they disappear, so the agent only sees the actions of the current screen.
Human-in-the-Loop sandbox. High-risk operations (e.g. processing payments, clearing cart) require explicit user approval.
Stream PCM audio packages directly between the browser, OwlLayer Server, and LLM providers for natural voice interactions.
Out-of-the-box integrations for React, Vue, Svelte, Angular, Vanilla Browser, Shopify, and WooCommerce.
The user says what they want, hands free. The agent built into the shop searches, filters and opens the product with the tools each screen declares, then carries on to the checkout and asks the user before paying.
Allow vocal or text agents to filter products catalog, update shopping carts, apply coupon codes, and guide customers directly to checkout.
Help users navigate complex dashboards, pre-fill tedious profile address form fields, and resolve settings options using simple speech.
Command complex analytical visual charts, adjust time range sliders, filter metrics categories, and export records instantly.
Allow workers to log notes, query checklists, and execute commands via high-fidelity audio streams while working on-site.
Register a local function inside your front-end component using React hooks. The tool lives and dies with the component lifecycle.
import { useAgentTool } from '@owllayer/react';
import { z } from 'zod';
useAgentTool({
name: 'apply_promo_coupon',
description: 'Apply checkout discount code',
schema: z.object({ code: z.string() }),
risk: 'low'
}, async ({ code }) => {
const result = await cart.apply(code);
return { success: result.ok, discount: result.amount };
});Create a lightweight OwlLayer Server orchestrator powered by Gemini or GPT adapters.
import { OwlLayerServer } from '@owllayer/server';
import { GoogleAdapter } from '@owllayer/adapter-google';
const server = new OwlLayerServer({
llm: new GoogleAdapter({ apiKey: process.env.GOOGLE_API_KEY }),
port: 4001
});
server.listen();Embed the UI widget block or talk to the client. When you type or say "Apply coupon SALE10", the server coordinates the LLM, validates parameters via Zod, forwards the execution call to your React handler, and streams back the result!