UnifiedLayer

🔌

OpenAI-Compatible API

Drop UnifiedLayer into any existing OpenAI integration. Use POST /api/v1/chat/completions with the same request/response format you already know, plus an unifiedlayer extension for sources, credits, and analytics.

🎨

Generative UI

Let the AI render React components — not just text. Register components with defineComponent(), and the model will call render_ComponentName with AI-generated props whenever a visual response is more appropriate.

🛠️

Tool Calling

Give the AI access to real-world actions. Define tools with defineTool() and the AI will call them mid-conversation, receive results, and continue — all automatically managed by the SDK.

🧵

Thread Management & History

Every conversation lives in a persistent thread. The SDK auto-creates threads, loads history on mount, and sends messages via POST /api/v1/threads/:id/advance — so users always pick up where they left off.

📚

Knowledge Base (RAG)

Upload PDFs, Markdown files, and plain text to a Knowledge Base in your dashboard. Documents are chunked, embedded via Gemini, and stored in Pinecone — then attached to your app for automatic context retrieval on every query.

🔐

Built-in Auth

Two auth modes for every use case. Use App-key headers for secure server-to-server calls, or create per-user session tokens (ul_sess_xxx) for safe browser-side integrations without exposing your API key.

Quick Start

Add a full AI chat interface to your Next.js app in under 10 lines:

bash

npx create-unifiedlayer-app

Or set it up manually:

tsx

// app/page.tsx
import { UnifiedLayerProvider, UnifiedLayerChatUI } from 'unifiedlayer-sdk'

export default function Home() {
  return (
    <UnifiedLayerProvider
      config={{
        baseUrl: process.env.NEXT_PUBLIC_UL_BASE_URL!,
        clientId: process.env.NEXT_PUBLIC_CLIENT_ID!,
        appId: process.env.NEXT_PUBLIC_APP_ID!,
        apiKey: process.env.NEXT_PUBLIC_API_KEY!,
      }}
    >
      <UnifiedLayerChatUI />
    </UnifiedLayerProvider>
  )
}

That's it. UnifiedLayer handles session creation, thread management, message streaming, and rendering — including generative UI components and tool calls.

How It Works

Your App (React)
     │
     │  UnifiedLayerProvider (SDK)
     │    ├── useUnifiedLayerSession   → POST /api/v1/auth/session   → session token
     │    ├── useUnifiedLayerThread    → POST /api/v1/threads        → thread created
     │    └── useUnifiedLayerThreadInput → POST /api/v1/threads/:id/advance
     │
     ▼
UnifiedLayer Platform
     ├── RAG retrieval  (Pinecone + Gemini embeddings)
     ├── LLM inference  (Groq — llama-4-maverick-17b)
     ├── Tool execution loop
     └── Response (OpenAI format + unifiedlayer extension)

Explore the Docs

Section	Description
Getting Started	5-minute setup with `npx create-unifiedlayer-app` or manual install
Core Concepts	Apps, Clients, Threads, Sessions, Tools, Generative UI, RAG
React SDK	UnifiedLayerProvider, hooks, defineComponent, defineTool
API Reference	All REST endpoints with request/response schemas
Dashboard	Manage Apps, Knowledge Bases, and Analytics

Section

Description

Getting Started

5-minute setup with npx create-unifiedlayer-app or manual install

Core Concepts

Apps, Clients, Threads, Sessions, Tools, Generative UI, RAG

React SDK

UnifiedLayerProvider, hooks, defineComponent, defineTool

API Reference

All REST endpoints with request/response schemas

Dashboard

Manage Apps, Knowledge Bases, and Analytics