⚡ ZeroRoute SaaS

1-Line Custom AI Chatbot & OpenAI-Compatible Multi-Cloud API Gateway

Zero Cost. Max Route. 100% Uptime Across 12 Pooled AI Cloud Providers.

Live SaaS • Subscriber Console • Master Admin • Buy Me a Coffee

---

🌟 What is ZeroRoute?

ZeroRoute is a production-ready, dual-purpose AI platform built on Next.js 16 (Turbopack), React 19, and Tailwind CSS v4:

1. 1-Line Embeddable AI Chatbot: Add a custom AI customer assistant to any website in under 60 seconds. Supports automated website scraping, Notion/Google Docs syncing, custom personas, brand colors, and streaming token responses.
2. OpenAI-Compatible Multi-Cloud Gateway: An ultra-fast drop-in replacement for OpenAI API endpoints (https://zeroroute.mapki.in/v1) that intelligently pools free tier quotas across 11 major AI clouds with sub-8ms auto-failover, zero rate-limit queuing, and 0ms RAM caching.

---

✨ Core Features

💬 1. 1-Line Embeddable Chatbot

• Embed anywhere with a single script tag before </body>:

  <script
    src="https://zeroroute.mapki.in/widget.js"
    data-bot-id="bot_your_bot_id_here"
    defer
  ></script>

• Live SSE Streaming: Fluid, token-by-token typewriter effect.
• Visitor Question Logs & Insights: Real-time stream of all visitor questions and AI answers in your dashboard so you finally know what people actually want from your site.
• Customization: Configure bot name, role, tone, greeting, avatar, brand accent colors, and quick prompt pills.
• Domain Whitelisting (CORS): Restrict widget execution to authorized customer domains.
• White-Label Branding: Pro subscribers enjoy 100% clean white-labeling with no external badge.

🧠 2. Semantic Document RAG & URL Crawler

• 1-Click Live Web Scraper: Crawl websites, Google Docs (/pub), Notion public pages, and GitHub raw Markdown (.md) with built-in SSRF protection.
• Context Injection: Intelligently retrieves verified facts and injects them directly into system prompts.
• Zero Hallucination Guardrails: Strictly instructs the model to refuse off-topic inquiries or fabricated information.

🌐 3. 11-Provider Multi-Cloud Gateway Pool

• 11 Configured Providers: Groq, Cerebras, SambaNova, Mistral AI, Google Gemini, NVIDIA NIM, Cohere, OpenRouter, Cloudflare Workers AI, Hugging Face, and BazaarLink AI.
• Sub-8ms Auto-Failover: Instantly reroutes queries to the next healthy provider if rate-limited (HTTP 429) or during provider downtime (HTTP 5xx).
• 0ms RAM Caching: Identical prompts are served instantly from RAM without consuming provider tokens.

💳 4. Turnkey SaaS Billing & Subscriber Console

• Subscriber Dashboard (/app): Manage API keys, customize chatbot persona, index knowledge bases, view Visitor Question Logs, search questions, and monitor monthly requests.
• Pro Tier ($2.00 / month): 10,000 monthly requests (~330 req/day), white-label widget, 3 domain whitelists, and priority routing.
• Free Tier ($0 / month): 500 monthly requests with automatic 30-day billing cycle resets.
• Dodo Payments Integration: Webhook synchronization with HMAC-SHA256 signature verification and replay protection.

👑 5. Master Admin Console (/admin)

• Multi-Cloud Benchmark Race: Test and measure live latencies across all 11 clouds in parallel with 1 click.
• Model Catalog Explorer: Browse and configure supported models per provider.
• Live Traffic & Failover Log Inspector: Real-time audit logs of requests, latencies, tokens, and multi-cloud fallback trails.
• Tenant Directory: Manage subscribers, quotas, and API keys.

---

🏗️ Gateway Multi-Cloud Hierarchy

flowchart TD
    Client["Client / Widget / OpenAI SDK"] --> Gateway["ZeroRoute API Gateway (/v1/chat/completions)"]
    
    Gateway --> RAG["RAG Retrieval Engine"]
    RAG --> Cache{"RAM Cache"}
    
    Cache -->|"Cache Hit (0ms)"| Immediate["Instant Cached Response"]
    Cache -->|"Cache Miss"| ProviderPool["Dynamic Multi-Cloud Pool"]

    subgraph Matrix["12 Pooled AI Clouds"]
        P1["1. Groq — ~100ms (Llama 3.3 / GPT-OSS)"]
        P2["2. Cerebras — ~80ms (Wafer-Scale 70B)"]
        P3["3. SambaNova — ~360ms (Gemma 4 / MiniMax)"]
        P4["4. Mistral AI — ~390ms (Mistral Small / Nemo)"]
        P5["5. Google Gemini — ~710ms (Flash 2.0 / 2.5)"]
        P6["6. NVIDIA NIM — ~260ms (Nemotron 30B)"]
        P7["7. Cohere — Conversational Reasoning"]
        P8["8. OpenRouter — Free Model Pool"]
        P9["9. Cloudflare AI — Edge Llama 3.1"]
        P10["10. Hugging Face — Serverless Router"]
        P11["11. BazaarLink AI — Qwen 2.5"]
        P12["12. IFM AI — Cerebras 375B MoE & B200"]
    end

    ProviderPool --> P1
    P1 -->|"Failover"| P2
    P2 -->|"Failover"| P3
    P3 -->|"Failover"| P4
    P4 -->|"Failover"| P5
    P5 -->|"Failover"| P6
    P6 -->|"Failover"| P7
    P7 -->|"Failover"| P8
    P8 -->|"Failover"| P9
    P9 -->|"Failover"| P10
    P10 -->|"Failover"| P11
    P11 -->|"Failover"| P12

Loading

---

⚡ Quickstart & Local Setup

1. Clone & Install

git clone https://github.com/amjadlle/zeroroute.git
cd zeroroute
npm install

2. Configure Environment

cp .env.example .env.local

Fill in .env.local:

APP_URL=http://localhost:3000
ADMIN_EMAIL=mapkisolutions@gmail.com
ADMIN_PASSWORD=your_secure_password
ADMIN_KEY=zr_admin_master_secret
ROUTER_API_KEY=zr_admin_master_secret

# (Optional) Cloud Provider API Keys
GROQ_API_KEY=gsk_...
GEMINI_API_KEY=...
MISTRAL_API_KEY=...
CEREBRAS_API_KEY=...

3. Run Development Server

npm run dev

• Landing Page: http://localhost:3000
• Subscriber Console: http://localhost:3000/app
• Master Admin: http://localhost:3000/admin

---

💻 API Gateway Usage

Python (Official OpenAI SDK)

from openai import OpenAI

client = OpenAI(
    base_url="https://zeroroute.mapki.in/v1",
    api_key="zr_live_YOUR_API_KEY"
)

response = client.chat.completions.create(
    model="auto", # Automatically routes across 11 clouds
    messages=[
        {"role": "user", "content": "How does ZeroRoute multi-cloud auto-failover work?"}
    ]
)

print(response.choices[0].message.content)

TypeScript / JavaScript (Official OpenAI SDK)

import OpenAI from "openai";

const openai = new OpenAI({
  baseURL: "https://zeroroute.mapki.in/v1",
  apiKey: "zr_live_YOUR_API_KEY",
});

async function main() {
  const stream = await openai.chat.completions.create({
    model: "auto",
    messages: [{ role: "user", content: "Tell me about quantum computing." }],
    stream: true,
  });

  for await (const chunk of stream) {
    process.stdout.write(chunk.choices[0]?.delta?.content || "");
  }
}

main();

---

🔒 Security Architecture

• Next.js 16 Edge Proxy (proxy.ts): Edge route protection on administrative and subscriber endpoints.
• Atomic Quota Counters: Prevents concurrent request quota race conditions.
• Cryptographic CSPRNG: Safe generation of session tokens and OTPs with crypto.randomInt and crypto.randomBytes.
• SSRF Hardened Web Scraper: Blocks private IPv4/IPv6, loopback, and metadata network endpoints.
• Standard HTTP Security Headers: HSTS, X-Frame-Options: SAMEORIGIN, X-Content-Type-Options: nosniff.

---

👨‍💻 Creator & Maintainer

Amjad P A

• Website: amjad.mapki.in
• GitHub: @amjadlle
• X / Twitter: @amjadlle
• LinkedIn: linkedin.com/in/amjadlle
• Buy Me a Coffee: buymeacoffee.com/amjadlle

---

📄 License

This project is licensed under the MIT License.