Introduction
Two ways to build an agentic application
Pick your track
Build with AI
Start from nothing. Describe your business and a real AI agent builds the application in front of you: your brand, your real data, your screens. Publish with one click and your website and MCP app are live. No code required.
Develop
One React component, and instead of plain text your users get branded, interactive screens generated live, per user, per question. Which agent answers is your call: the agent BranderUX hosts for your project, with no backend at all, or the agent you already have, which keeps the reasoning and fills the screens on your side.
The Develop track
Everything below covers the Develop track end to end. With no handler the project's hosted agent answers, so the first integration is one component. With your own agent, any AI provider or agent framework works, from LangGraph, CrewAI, Google ADK, Pydantic AI, Mastra, and AutoGen to direct SDKs like Anthropic, OpenAI, and Gemini, and the Vercel AI SDK.
Key Features
Drop a single React component into your app
Progressive streaming UI, components load in real-time as the AI responds
Works with any AI, LangGraph, CrewAI, Anthropic, OpenAI, Gemini, and more
AG-UI Protocol support, standard streaming events for modern agent frameworks
Fully interactive, every element generates new queries
Automatically styled to match your brand
How It Works
Every answer becomes a screen. Here is the loop your users live in:
- 1. User types a query
e.g., "Show me sales dashboard"
- 2. BranderUX selects the best screen layout
Analyzes the query and picks from dashboard, table, chart, etc.
- 3. Your AI streams structured data
Using AG-UI events for real-time progressive loading
- 4. UI components appear as data streams in
Charts, tables, stats, all rendering progressively
- 5. Every interaction generates a new query
Click a row, filter a chart, the cycle continues seamlessly
Simple Example
import Brander from "@brander/sdk";
function App() {
return <Brander apiKey="bux_pk_your_key" projectId="your_project_id" />;
}Or pass a stream handler and the agent you already have fills the screens:
import Brander, { sseStream } from "@brander/sdk";
function App() {
return (
<Brander
apiKey="bux_pk_your_key"
projectId="your_project_id"
onQueryStream={(params) => sseStream("/api/agent", { params })}
/>
);
}Your backend returns Server-Sent Events (SSE) with AG-UI events. Most agent frameworks support this natively.
Supported Agent Frameworks
BranderUX works with all major agent frameworks through the AG-UI protocol:
- LangGraph, Most popular
- CrewAI, Multi-agent
- Google ADK, Gemini ecosystem
- Pydantic AI, Type-safe Python
- Mastra, TypeScript native
- AutoGen/AG2, Microsoft
For direct SDK usage without a framework, we provide adapters for Anthropic, OpenAI, and Gemini.
Direct SDK Example
If you're calling an AI provider directly (without an agent framework), use our stream adapters:
import Brander, { anthropicStream } from "@brander/sdk";
import Anthropic from "@anthropic-ai/sdk";
const anthropic = new Anthropic({ apiKey: process.env.ANTHROPIC_API_KEY });
function App() {
return (
<Brander
apiKey="bux_pk_your_key"
projectId="your_project_id"
onQueryStream={async function*(params) {
const stream = anthropic.messages.stream({
model: "claude-sonnet-5",
// Your persona + BranderUX UI instructions, append, never replace
system: YOUR_SYSTEM_PROMPT + "\n\n" + params.system,
messages: params.messages,
tools: params.tools?.anthropic,
max_tokens: params.max_tokens || 4000,
});
yield* anthropicStream(stream);
}}
/>
);
}Next Steps
- Installation, Get @brander/sdk installed in your project
- Quick Start, Create your first widget in 5 minutes
- Agent Examples, Full examples for LangGraph, CrewAI, and more
- Props Reference, Complete API documentation