After weeks of experimenting, I finally built an AI support assistant that can train itself using a company's own website and knowledge base.
Instead of manually writing hundreds of FAQs, the chatbot crawls website pages and answers customer questions using that content.
What it can do
✅ Crawl website pages and build a knowledge base
✅ Answer product, shipping, return, and policy questions
✅ Reduce repetitive customer support tickets
✅ Works 24/7
✅ No-code setup
I recently created a demo using an eCommerce fashion store, and it answered questions based on the website content instead of using generic AI responses.
One thing I learned while building this project:
The quality of an AI chatbot depends far more on the quality of its knowledge base than on the LLM itself.
Next on my roadmap:
- Better website crawling
- Automatic re-training
- Shopify integration
- Analytics dashboard
- Human handoff
- Multi-language support
I'd love feedback from developers:
If you were building an AI customer support tool today, what feature would you consider essential?
Demo:
https://support-ai-3bba5c.netlify.app/
Top comments (0)