🧠 KnowBase AI

Build custom AI chatbots trained on your documents, URLs, and Q&A data — with citations, embed widgets, and real-time conversations. A production-ready, self-hostable Next.js SaaS boilerplate with semantic context retrieval and iframe-embeddable bots. A free open-source alternative to Chatbase, CustomGPT, Botpress, SiteGPT, and Stack AI — powered by the MuAPI AI engine.

Tech stack: Next.js 14 (App Router) · Prisma · PostgreSQL · NextAuth (Google OAuth) · Stripe · Tailwind CSS · MuAPI · RAG / semantic retrieval · OpenAI Use cases: Customer support automation · Internal documentation Q&A · Sales lead qualification · Documentation chatbots · Course/training assistants · SaaS help bots · Knowledge management · Onboarding assistants · Embeddable site chat widgets

🎨 Explore 50+ more open-source AI apps →

🌐 Project Details

GitHub Repository: github.com/SamurAIGPT/ai-knowledge-base

Live Demo Preview: ai-knowledge-base-six.vercel.app

📸 Studio Workspace Preview

---

KnowBase AI is a production-ready, highly-optimized AI web application. Out of the box, it seamlessly manages User Authentication, Credits & Billing, Custom Document Persistence, semantic keyword-matched context retrieval, and AI generation using Next.js (App Router) architecture. It empowers businesses, developers, and creators to build custom chatbots with specific datasets.

Why use KnowBase AI?

• Production-Ready SaaS — Complete with Google OAuth and Stripe Checkout workflows built-in.
• Custom Knowledge Builder — Create sandbox bases, load web URLs, custom text files, or write Q&A entries.
• High-Contrast Playground — Chat dynamically with custom bots showing exact reference matches and citation sources below dialogue bubbles.
• Embeddable Widgets & API Access — Generate iframe scripts to embed customized chatbots directly on third-party blogs or landing pages.
• Responsive Screen-Fitting — Dynamic viewports stacked as clean scrolling components on mobile and locks screen height on desktop for premium desktop UX.

---

✨ Core Features

🎨 Custom Knowledge Sandboxes (Main Page /)

• Sandboxes list sidebar with modal trigger to forge new custom RAG spaces.
• Cost: Free (0 credits) per Knowledge Base created.

🗄️ Multi-Source Trainer

• Scrap URL: Submit document links, scrap site contents, and index automatically.
• Add Q&A: Custom type precise matching prompts and answers.
• Upload File: Insert custom raw documents data.
• Cost: 10 credits per source trained.

💬 Dialogue Playground

• Natural dialogue balloon interface with real-time semantic context injection.
• Expandable citation logs referencing matching sources and snippet previews.
• Cost: 2 credits per chatbot query.

💳 Stripe Credit Billing (/pricing)

• Four credit packs based on a $1 = 200 credits conversion rate:
  • Basic Pack ($5 / 1,000 credits)
  • Standard Pack ($10 / 2,000 credits)
  • Professional Pack ($20 / 4,000 credits — Most Popular)
  • Business Pack ($50 / 10,000 credits)
• Credit balance is automatically topped up via Stripe webhook on checkout completion.

---

⚡ Deployment: Vercel & Production

This architecture is engineered explicitly for Vercel serverless environments.

One-Click Deploy

Live App: ai-knowledge-base-six.vercel.app

🔑 Required Environment Variables

To successfully deploy and run, you must populate the following environment variables in your Vercel project settings:

Service	Variable	Description & Source
Database	DATABASE_URL	PostgreSQL connection string (Supabase or Neon)
NextAuth / Google	NEXTAUTH_SECRET	Secure random string generated via openssl rand -base64 32
	NEXTAUTH_URL	Your production domain (e.g. https://my-app.vercel.app)
	WEBHOOK_URL	Public URL for MuAPI async callbacks (same as NEXTAUTH_URL in production)
	GOOGLE_CLIENT_ID	Get from Google Cloud Console
	GOOGLE_CLIENT_SECRET	Get from Google Cloud Console
Stripe Billing	STRIPE_SECRET_KEY	Get from Stripe Dashboard
	NEXT_PUBLIC_STRIPE_PUBLISHABLE_KEY	Get from Stripe Dashboard
	STRIPE_WEBHOOK_SECRET	Webhook secret for resolving credit purchases
AI LLM Queries	MU_API_KEY	Create an account and get key from muapi.ai/access-keys

🚀 Launching on Vercel: Step-by-Step

1. Database Provisioning: Create a new Postgres database (via Supabase or Neon). Retrieve the connection string (DATABASE_URL).
2. Project Creation: Import your GitHub fork into the Vercel dashboard.
3. Configure Environment Variables: Copy the variables above into the Vercel project settings environment tab.
4. Deploy: Hit "Deploy". Vercel will automatically run the build steps (npm run build).
5. Database Push: Run npx prisma db push to synchronize database models before launching.
6. Integrations Setup:
   • Establish a Google Cloud OAuth app, enabling the callback URL: https://your-app.vercel.app/api/auth/callback/google
   • Setup a Stripe Webhook, pointing to https://your-app.vercel.app/api/stripe/webhook and selecting the checkout.session.completed event.

---

🛠️ Local Development

Ready to iterate locally? Setup is straightforward.

Prerequisites

• Node.js (v18 or higher)
• A local PostgreSQL instance or a free cloud Database URL.

Setup

# 1. Clone the repository
git clone https://github.com/SamurAIGPT/ai-knowledge-base
cd ai-knowledge-base

# 2. Install dependencies
npm install

# 3. Setup Environment
cp .env.example .env
# Open .env and insert your specific keys.

# 4. Initialize Database Schema
# Note: Because the database is shared, see the Safety Warning below!
npx prisma generate
npx prisma db push

# 5. Start the Development Server
npm run dev

The console should now be active on http://localhost:3000.

---

⚠️ Database Safety Warning (Shared Pool)

The workspace database is shared with other applications. Running npx prisma db push on a clean, empty schema will drop tables belonging to other applications. Always follow the Pull-Declare-Push-Cleanup sequence:

1. Run npx prisma db pull to fetch all database tables.
2. Declare your KnowledgeBase, Source, KBChat, and KBMessage tables and update the relations on the User model.
3. Run npx prisma db push to add your changes safely.
4. Clean up schema.prisma to keep only NextAuth models, KnowledgeBase, Source, KBChat, KBMessage, and the updated User relations.
5. Run npx prisma generate to rebuild the type-safe client.

---

🏗️ Technical Architecture

ai-knowledge-base/
├── prisma/
│   └── schema.prisma           # Postgres schema (User, Account, Session, KnowledgeBase, Source, KBChat, KBMessage)
├── src/
│   ├── app/                    # Next.js App Router
│   │   ├── page.js             # Main Workspace Studio (Sources, Playground RAG, Widget export)
│   │   ├── dashboard/          # Personal gallery list overview
│   │   ├── pricing/            # 4-Plan credit pricing grid structured at \$1 = 200 credits
│   │   └── api/
│   │       ├── auth/           # NextAuth dynamic sessions
│   │       ├── kb/             # GET / POST / DELETE knowledge bases
│   │       ├── kb/[id]/sources # GET / POST / DELETE data sources (Files, URLs, Q&As)
│   │       ├── kb/[id]/chat    # GET / POST / DELETE playground chats
│   │       ├── kb/[id]/chat/[chatId]/messages # GET / POST query messages (Context, citations, LLM)
│   │       └── stripe/         # Stripe checkout creation + checkout webhook
│   ├── components/
│   │   ├── Providers.js        # NextAuth SessionProvider wrapper
│   │   └── layout/Navbar.jsx   # Sticky header with Vercel Deploy button & credit balance
│   └── lib/
│       ├── auth.js             # NextAuth config with Prisma adapter
│       ├── config.js           # Central config mapping Google, Stripe, MuAPI keys
│       ├── prisma.js           # Cached Prisma client singleton
│       ├── stripe.js           # Stripe instance initializer
│       └── services/
│           ├── user.js         # Credit management service
│           └── billing.js      # Stripe checkout and payment webhook parser
└── next.config.mjs             # Next.js configuration

---

📄 License

MIT Licensed.

---

KnowBase AI: A premium, high-contrast, fully responsive customized semantic knowledge agent SaaS builder.