Is it possible to make chatbot answer only based on trained data from pdf embedding? Like not showing users 'create Index and Embedding' button or uploader, not using flowise or integrate flowise with nextjs. I appreciate if you make tutorial of it
Of course it is possible. I added the upload function as a bonus so you can just delete it from the code if necessary and you have the “user” version. And keep the original as the “admin” version where you can add information ℹ️ you want them to search
Great video !! I really like to see more code videos from you. For the create embedding and index process can we automize it so it will automatically create new embeddings and indexes if we added a new file into the documents folder?
Sure, just depends where the folder is? If somewhere in the cloud you can probably use that hostings api to fire 🔥 when new file is added and pass information to the embedder database.
Stack AI is a bit expensive per license Plus botpress together would cost about $225/month (around 5000 messages). Will make a review if both platforms in the next video
Well Supabase records it in the regular table, nothing fancy. I saw a doc about full text search using Supabase supabase.com/docs/guides/database/full-text-search maybe try that. What are you building?
you are awsome my man. keep up the good work. love your vids!
Thanks man ☺️ this ☝️ is not popular but I feel like has the most value
Super clean code and explanation !
Thank you 🙏🏻 I spent a lot on that video
Is it possible to make chatbot answer only based on trained data from pdf embedding? Like not showing users 'create Index and Embedding' button or uploader, not using flowise or integrate flowise with nextjs. I appreciate if you make tutorial of it
Of course it is possible. I added the upload function as a bonus so you can just delete it from the code if necessary and you have the “user” version. And keep the original as the “admin” version where you can add information ℹ️ you want them to search
Yes it is possible, just add it in the prompt if no return to the vector search say “ I don’t know “
Great video !! I really like to see more code videos from you.
For the create embedding and index process can we automize it so it will automatically create new embeddings and indexes if we added a new file into the documents folder?
Sure, just depends where the folder is? If somewhere in the cloud you can probably use that hostings api to fire 🔥 when new file is added and pass information to the embedder database.
Great videos! I would love to hear your input on botpress + stack AI for creating chatbots!
Stack AI is a bit expensive per license Plus botpress together would cost about $225/month (around 5000 messages). Will make a review if both platforms in the next video
How can we integrate search to this chatbot? Also is this going to work for multi tenant architecture?
Well Supabase records it in the regular table, nothing fancy.
I saw a doc about full text search using Supabase supabase.com/docs/guides/database/full-text-search maybe try that. What are you building?
Show me more 🎉❤🎉
Next one will be much lighter video on how to use bubble.io make a personal Chatbot 🤌