AssistFlow

Inspiration

Business often struggles with slow customer support, repetitive queries, high workload, and inconsistent responses, along with need for 24/7 availability.

What it does

I build an AI support system that turns company documents into an intellegent helpdesk. After signin User get an tag, that they can paste into any HTML file directly. Thus, Customer support chatbot starts working.</p> <h2 id="how-we-built-it">How we built it</h2> <p>We build it using python and fastapi. Databases include SQL, Vector, and Redis to store and cache client data. </p> <h2 id="challenges-we-ran-into">Challenges we ran into</h2> <p>During Development we find out some queries need escalation to a real human. Thus we add an escalation system to inform user about the complaints via email.</p> <h2 id="accomplishments-that-were-proud-of">Accomplishments that we&#39;re proud of</h2> <p>Created a true zero-integration chatbot that businesses go live in under a minute by pasting just one script tag and built a production-ready RAG pipeline with smart, content-aware chunking that delivers accurate, grounded answers from real company documents (including code files).</p> <h2 id="what-we-learned">What we learned</h2> <p>We learned that chunking strategy beats model size. Smaller models can perform well with better chunking.</p> <h2 id="whats-next-for-assistflow">What&#39;s next for AssistFlow</h2> <p>We will expand intent detection to automatically handle buying signals, demo requests, and contact needs with appropriate CTAs or calendar booking links.</p>

Built With

• api
• css
• fastapi
• html
• javascript
• python
• rag
• redis
• render
• resend
• sql
• vector