Qunivex

Create AI ChatBot for your website with just one <scrip> tag

August 29, 2026 I built a website named Qunivex, so anyone can embed a ChatBot in their website with just a single <script> tag.

So I've been working on a project called Qunivex for a long time now, and wanted to share about it to you all. It’s basically an Agentic AI builder designed to let anyone create highly customized AI chatbots and embed them onto any website using a single <script> tag.

The goal was to move past basic QA bots and make it easy to deploy actual "agents." Inside the Visual Studio, you can configure system prompts, connect knowledge bases, and chain together sub-agents, tools, and strict guardrails. It supports multiple frontier and open models (including Llama, Qwen, DeepSeek, and Mistral), so you aren't locked into one provider.

A big focus for me was security. Unlike standard widgets that leak your API keys to the browser frontend, Qunivex architecture keeps everything safely on the server side.

I’ve also built a dedicated WordPress plugin to make setup even faster for non-devs.

The core tech is up and running, and I’d love to get this community's brutal feedback. What features are missing for your use cases? How is the UI flow in the video?

Check it out at qunivex.com and let me know what you think!

VanshChoyal

2 Likes

3 Comments

Say something nice…

Post Comment

1

The technical foundation is impressive, but I think the bigger product opportunity is after deployment.
A non-technical user can build and embed an agent, but how do they know it’s actually performing well once it’s live?
A simple layer for conversation quality, failed answers, unanswered questions, and improvement suggestions could turn Qunivex from an agent builder into an agent optimization platform.
That feedback loop could become especially valuable as users manage multiple agents or knowledge bases.
Curious if you’re thinking about the post-deployment experience yet

Tayyaba Sajjad

·
8 days ago
·
Reply
1. 1
  
  Hey Tayyaba! I hadn't thought about post-deployment in quite as much detail as you've mentioned. On Qunivex, you can already use the Playground to check response quality, and the Logs tab can help identify questions the agent failed to answer or other unusual queries. I'd definitely implement logic to automatically filter unanswerable questions, and potentially hand the conversation over to a real person if the client has the resources for that. I also checked out Appvion Studio, and I think Qunivex could actually be useful for your website. I've already trained an agent on the site, so if you have a few minutes, I'd genuinely love for you to try it and see how the responses hold up: https://qunivex.com/d/BHFAD0AD345GACVMQRRYZJ This is a preview page, so the UI and reasoning view aren't representative of the full Qunivex Dashboard. You can also claim the agent from the button below if you'd like to explore it further. Would love to hear what you think.
  
  VanshChoyal
  
  ·
  11 hours ago
  ·
  Reply
  1. 1
    
    I gave it a proper look, and the idea of automatically identifying unanswerable queries is particularly interesting. That’s where I’d be watching closely, because response quality alone doesn’t tell you whether an agent is actually reliable in production. I’ll test it from a few real user scenarios and edge cases and see how it holds up.
    
    Tayyaba Sajjad
    
    ·
    11 hours ago
    ·
    Reply

Qunivex

About

People