Is there any way to allow a user to upload attachments or docs to a pickaxe that can be used during the chat conversation, but not necessarily adding the document to the kb?
Thanks,
Graydon
Yes, there is! This is exactly what the end-user document upload feature doesn’t.
If you enable users to upload their own documents into a tool, then a paperclip icon appears on the tool and end-users can upload the documents.
End-user uploaded documents are added to the knowledge base of the tool for only that user’s particular session. Only that end-user in that session will be experiencing a tool with that document in it.
Here is a screenshot of where the toggle is for allowing end-user uploads.
1 Like
Does this apply to forms as well? So to be clear, if end users upload their documents or images in a chat or form, that won’t count towards the knowledge base document limit?
Also, for the form document upload, is there a way to customize the language “Drag and drop a PDF or…”
I would like to prompt the user to upload an image instead of a PDF and don’t see a way to change it currently. Could it be possible (in forms, not chat) to make the document/image upload optional as well?
Thanks!
Yes, you can enable end-user document upload for both chatbots and smart-forms. These uploads do not use credits nor do they count against your knowledge base limit. However, I believe end-users cannot upload gigantic documents. The 100mb limit applies to them. For text-based files this is fairly massive.
This text is not customizable. You bring up a great point though! We just added vision models to the platform, so have overlooked that piece of copy in the end-user upload field in Smart Forms. Thanks for pointing this out. We will update that copy.
Great, thank you! One more note: it looks like the upload section becomes required for users if we use forms and include it. Could it be possible to make upload an optional field?
Yes for form based pickaxes all input fields (text or file upload) are mandatory. That’s just how it works currently.
Hey @admin_mike regarding documents the user uploads, can the user only interrogate the document? or is it possible for the user to ask the tool to display content from it? I’m trying to create a tool that improves and rewrites resumes, but I’m having difficulties getting the chatbot to display anything from the document uploaded
Cheers
Alex
Hey Alex, if the document is below a certain size (like a resume) the entire contents of it is dumped into the context window of the model. In contrast, if they upload a 300 page book, it is handled the normal way.
So in your example, the bot should be able to read and duplicate the material very easily.
1 Like
thank you @admin_mike the issue was on the prompting, all sorted now, cheers
1 Like