ServU is a platform that lets anyone drop an intelligent AI chatbot onto their website with just a few lines of code—or even just a CDN script tag. No AI expertise required, no complex setup, just install an NPM package or copy-paste some JavaScript and you're live. Behind the scenes, it uses RAG (Retrieval Augmented Generation) with vector databases to make chatbots that actually know your business, support ticketing to escalate to real humans when needed, and Groq for lightning-fast responses that don't make users wait.

Every business wants AI chatbots now. The problem? Most solutions require you to be a developer, understand embeddings and vector databases, set up your own infrastructure, and somehow integrate it all with your website. For low-code founders or business owners who just want to improve their customer support, that's way too much friction.

And even when you do get a chatbot running, most of them give generic responses that frustrate customers more than they help. They don't know your products, your documentation, your specific business context. Customers end up waiting for human support anyway.

I wanted to build something that solved both problems: make it stupid easy to install, and make the bots actually useful by connecting them to your real business knowledge.

ServU gives you a complete chatbot platform without requiring any AI knowledge. You create and configure your chatbot through a web dashboard, then install it on your site with either an NPM package or a simple script tag. That's it.

Behind the scenes, it handles all the complex stuff: OpenAI embeddings for semantic understanding, vector database storage for your business knowledge, Groq for fast inference, and Supabase + AWS for reliable backend infrastructure. Your chatbot learns from the documentation and data you provide, so it actually gives relevant, informed answers instead of hallucinating nonsense.

When the bot can't help, it can create support tickets or refer customers to real humans. Everything is tracked so you can follow up on queries and continuously improve your support.

1. Create Your Bot: Log into the web dashboard, create a chatbot, and upload your business documentation, FAQs, or knowledge base
2. Configure Behavior: Use custom instructions to shape how the bot responds, enable ticketing features, customize colors and appearance
3. Knowledge Processing: The system creates OpenAI embeddings from your data and stores them in a vector database for semantic retrieval
4. Install on Your Site: Install via NPM (npm install @servu-ai/servu-chat) or just add a script tag for vanilla JavaScript
5. Lightning-Fast Responses: When users ask questions, the system retrieves relevant context and uses Groq for near-instant AI responses

The web dashboard is where you manage everything. Create new chatbots (we call them "butlers" because it sounds fancier), configure their personalities, and upload your business knowledge.

The training interface lets you upload documents, add website URLs to scrape, or manually input information. Everything gets processed into embeddings and stored in the vector database so your bot can retrieve relevant information when answering questions.

Customization options let you control the bot's appearance, behavior, and features. Change colors, adjust response style, enable ticketing, set business hours—all through a straightforward UI.

The whole point was making installation trivial. Once you've configured your chatbot in the dashboard, you get a unique Butler ID. Then you have two options:

npm install @servu-ai/servu-chat

import { ServuChat } from '@servu-ai/servu-chat'

function App() {
  return <ServuChat butlerId="YOUR_BUTLER_ID" />
}

<script src="https://cdn.servu.ai/servu-chat.js"></script>
<script>
  ServuChat.init({ butlerId: 'YOUR_BUTLER_ID' })
</script>

That's it. No configuration files, no API keys to manage, no infrastructure to set up. The complexity is hidden behind a simple interface.

The project is split into three main components: a React + TypeScript widget (built with Vite), a Next.js web application for the dashboard, and a landing site. Everything is deployed on Vercel with Supabase handling the database and AWS for cloud storage. The AI stack uses OpenAI for embeddings, vector databases for semantic search, and Groq for inference because it's ridiculously fast compared to standard OpenAI models.

The NPM package (@servu-ai/servu-chat) supports multiple frameworks (React, Vue, Svelte) plus vanilla JavaScript, so it works with basically any web stack. Check it out on NPM.

I finished this project a few months ago and got it deployed at servu.ai, but honestly never put much effort into marketing or growing it. It just sat there, working fine, but not really going anywhere.

Now with all the hype around agentic AI and the new capabilities for connecting LLMs to external services and platforms, I'm seeing an opportunity to revive this project. The original vision was chatbots that could answer questions and create tickets. But what if they could actually do things?

Imagine chatbots that don't just tell customers how to track an order, but actually query your e-commerce platform and provide the tracking info. Bots that can schedule appointments, process refunds, update account information, or trigger workflows in your CRM. The infrastructure is already there—it just needs the agentic layer.

The timing feels right to add these capabilities and make a real push to grow this into something bigger. The core product works, the installation is dead simple, and the AI landscape has evolved enough that the value proposition is clearer than ever. Time to dust this one off and see where it can go.