continuing-cyan•17mo ago
I have been looking for tutorial on this for days but can't find any proper solution.
Hi all! I need your help.
I'm trying to build a cahtbot that is capable to take upload files, process them, and then output files.
Currently I created a GPT that can accept Invoices as PDFs, understand them, and break them down onto a CSV file, then allow me to download the CSV file.
My approach was to create a chatbot that allows uploads, then create a openAI assitant API that acceps that upload, reviews it, and generates the output, which is sent again to the chabot and provide us with the output. I've seen some options trying to integrate a bucket or similar to temprarly store the file but as I'm not a coder, it was too difuclut for me,.
Any ideas for an easier better option? Thank you so much!
3 Replies
Based on your requirements, it seems like you're on the right track with Voiceflow. Here's a step-by-step guide to help you create a chatbot that can process uploaded files:
1. Create an AI Assistant in Voiceflow: In your workspace, click the 'New Assistant' button on the top right. Give your assistant a name and select 'Chat' as the agent type.
2. Add Data to Your Assistant: Go to the Knowledge Base in your assistant's settings. Click on 'Add Data Source' on the top right and upload your PDF files. Your AI Assistant will use these documents to answer questions via GPT. Note that there is a 10mb file limit per upload.
3. Preview Your Assistant's Responses: After uploading some initial documents, press 'Preview' on the top of the screen. This will allow you to ask the Knowledge Base questions and see what the responses are.
4. Deploy Your Assistant: Use Voiceflow's API or Integrations to deploy your assistant to a website.
As for the file processing part, you might need to use an external service to convert the PDF invoices into CSV files. Once the file is processed, you can send the output back to the chatbot.
Unfortunately, Voiceflow does not natively support file uploads in the chat interface. You might need to use a third-party service to handle the file upload process.
For the non-coder part, you might need some technical assistance to integrate all these parts together. Voiceflow is designed to be user-friendly for non-coders, but some advanced functionalities might require technical knowledge.
I hope this helps! If you have any more questions, feel free to ask.
Relevant links:
1. Voiceflow Documentation
2. Voiceflow API Documentation
rising-crimson•17mo ago
File upload in the chat window is mooted as a chat widget extension that is in beta. At the moment all you can do is upload a file to a site and not do anything with it. Like you, I see immense value in being able to analyse an uploaded, temporary, file - it is a much asked for function. As are scroll bars in the embedded agent and a copy function. The relationship between customer satisfaction and product attributes is often overlooked in tech. See Kano's model: https://www.ifm.eng.cam.ac.uk/research/dmg/tools-and-techniques/kano-model/#:~:text=Developed%20in%20the%2080's%20by,between%20essential%20and%20differentiating%20attributes.
continuing-cyanOP•17mo ago
Thanks Steve. There re some works around to be able to manipulate files and use openai api to analyse them, but at the moment it doesnt seem straight foraward. Hopefully we get something done soon.