No that's not what you should do
There, you should pass the User message
OpenAI should be aware of the file you uploaded to your assistant
It has nothing to do with Typebot in theory
Oh I see that it's part of their v2 API
But still, file retrieval should work fine in v1
right that's what I thought. I shouldn't have to pass in any additional functions. but when I send a message with typebot i get an message back from the assistant that no files are uploaded. Not sure how to debug.
here is the repsonse from typebot -
here is the response for the same assistant in the playground
and even when I switch to v1 in the playground it still works
Ok let me try it out again
Ok I can reproduce the issue 🤔
Thank you for reporting. I will fix it most likely today
Moving from free to paying customer NOW! 🙂
Took more time than expected, but I'm shipping all the v2 goodies including streaming with the bug fix! I'll most likely merge it tomorrow as it needs more testing for now
oh wow you are a champion! Thank you @Baptiste !
it's looking great @Baptiste ! will let you know if I find any issues!
Assistants continuam com erro:
após a primeira interação com o Assistant do GPT, ele gera esse erro a não é mais possivel enviar novas mensagens ao GPT
escrevo a mensagem no Typebot, e ela fica aguardando a resposta, que nunca chega
a segunda mensagem nunca chegou a thread do assistant
Assistants continue to have an error
after the first interaction with the GPT Assistant, it generates this error and it is no longer possible to send new messages to GPT
I write the message in Typebot, and it is left waiting for a response, which never arrives
the second message never reached the assistant's thread
Another thing also broke with the new update.
Before, the Ask Assistant provided two returns, the ThreadId and the Message.
I used the ThreadId to configure the GPT Thread, but now this return no longer exists and the control became poor.
This change ended up breaking my automations and I am without a solution so far.
Indeed, you don't have to save the Thread ID anymore, it does it automatically!
Can you try deleting the ThreadId response item?
Hey @Baptiste - I'm seeing an issue with long chats, the assistant starts to hallucinate. Is there a way to limit the number of messages that are appended to the chat message to say the last 5 messages? The flow that I'm using is essentially your example out of the box -
Assistant API is using OpenAI built-in memory so your Assistant Message is not in use here, you don't need it
If the assistant is starting to hallucinate then the issue comes from OpenAI 🙁
helpful. I will look at the prompt then!
and just FYI @Baptiste in the Mesage block, I had created an instruction that looks like this:
Here is the current lecture that the user is watching: {{Video Name}}
Keeping your answers focused on {{Video Name}} please answer {{User Message}}.
Only answer questions about other lectures if the user specifically asks for it.
For context, here is the chat history {{Chat history}}
I believe that {{Chat history}} was creating a problem as I can't seem to reproduce now that I have taken that last bit out
