Verbel

• 

  Home page

Inspiration

Having first hand experience of studying online made us realise all the unused potential of recorded lectures, or any form of online educational content. Studying for an exam one had to check through all lectures and notes by hand, skipping through 90 minute videos. There just had to be a more intelligent way of accessing all this knowledge.

What it does

Verbel is a web app that allows users to upload their content such as pdfs, podcasts, and videos. This content is then made accessible through semantic search, which allows users to easily find the information they are looking for by asking natural language questions. The aim of Verbel is to provide users with a comprehensive knowledge base that is easy to use and navigate.

How we built it

• We used AssemblyAI for the video transcripts, cohere for embeddings and pinecone to store and query the embeddings.
• The frontend is built with Next.js and hosted on Vercel.
• Supabase was used for database and storage and Azure Container Instances for the API

What we learned

• How to manage multiple micro services
• Quickly build a usable MVP
• Combining multiple AI services to build a product

What's next for Verbel

• Additional formats (ebooks, podcasts, Notion, ...)
• Multiple features based on LLMs (summary, question answering, automatic study cards)

Built With

• assemblyai
• azure
• cohere
• docker
• nextjs
• pinecone
• python
• react
• supabase
• tailwindcss
• typescript