Integrate OpenAI API Assistants In Voiceflow (OpenAI DevDay Update)
Vložit
- čas přidán 28. 08. 2024
- How to integrate OpenAI API Assistants in Voiceflow. This API Voiceflow integration allows you to integrate API assistants in Voiceflow. The new OpenAI DevDay updates have brought many opportunities, and this is a big one. The full Voiceflow template is available down below.
Get the Free Template & community here: www.skool.com/...
Voiceflow: partners.voice...
Book your consultation call with us: calendly.com/o...
Our Links 🔗
Agency: omnifusion.ai
Instagram: / omnifusion.ai
#openai #voiceflow #aiautomationagency
Get the Free Template & community here: www.skool.com/omnifusion
Our Links 🔗
Agency: omnifusion.ai
Instagram: instagram.com/omnifusion.ai
Book your consultation call with us: calendly.com/omnifusion-consulting
If the resource is free, why do I have to wait for Admin's approval?
Thanks. I was able to get it to work. Of the many videos I've watched, even from VoiceFlow, none made it this simple. The template was easy to follow. Good job.
Glad it helped!
Amazing solution! Really nice that you have been so fast to address the issue of sharing custom gpt assistants with non-premium chatgpt users. There is an enormous value in this. You are helping me a lot to develop my AI agency. Btw, I cannot find the template in the resources hub.
Thank you for your kind words, really glad we could help! Giving value is what we're about.
And my apologies, forgot to publish it :D It's in the resource hub now!
thank you so much this was very helpful for my small team in argentina
Glad it helped!
with this approach, I don't have to use replit anymore? that's great.
Yep
Thanks for this. Finally found a straight forward to the point solution! Just wanted to check-in and see if anyone has suggestions for any modifications that might have come up with recent developments form OpenAi or Voiceflow. I thought about using the Knowledge feature in Voiceflow, but I am not sure there is a benefit to it if I cam already using retrieval in my openai assistant. GPT 3.5 did get a lot cheaper which is great. Thanks!
Welcome! We are launching our community today where you'll be able to discuss topics like these with others in the community. Join our resource hub to be notified when it drops. You definitely need to add time delays since the current version will make you hit the 60 RPM rate limit for assistants api quickly.
@OmniFusionAI ... A question.
Imagine you create a GPT that works like a doctor. In the GPT, you upload a file and the GPT gives you a diagnosis.
When I link the GPT with VOICEFLOW, how can we make our chatbot allow file uploads and be able to diagnose our injury?
We have not tested that yet, so unsure of how to help.
Thanks!
Welcome!
One more great video. You guys are really good.
A question.
Maybe I have configured something wrong, but using the OpenAI API is very expensive. I have done 5 very simple tests and they have cost me more than $1.
If we implement it on the web or on WhatsApp and it is used by users en masse, I am afraid that the cost will skyrocket...
Is this the case or am I doing something wrong?
Appreciate it! Glad you enjoy the value we put out :) As for the cost, you are probably using GPT-4? I would suggest GPT 3.5 1106, if you want to use retrieval with your API Assistant. In that case, costs should be minimal.
Fire video G, have a question here can I integrate open ai assistant to many chat? Am thinking about creating an instagram lead qualifying chatbot just like you did. Tell me which is the better way open ai assistant to voiceflow then to manychat or directly to manychat? thank you for providing great value. Hoping to working with you soon.
Appreciate it G! We use API Assistants --> Voiceflow --> Manychat. Reason being is that Voiceflow has a lot of functionality that we use for our projects and cannot be replicated with ManyChat. Including api calls, conditions, etc. So we haven't looked into doing it with ManyChat directly and probably won't look into that anytime soon.
If you have a VF subscription anyways, it doesn't matter. None of this uses any AI tokens inside of Voiceflow.
@@OmniFusionAI thank you reaching out back what you said really helps. appreciate you for posting the fire and valuable contents.
@@Kevin-pv6fg Sure thing, appreciate your support!
Thank you, appreciate your work, subbed.
Glad we could help!
thanks for the video , the only way to connect voiceflow to whatsapp is to create a Meta account and register your business? Is there any other way . Since I simply don't have time to wait 1 month to register my company.
Check this video: czcams.com/video/8cRdz0DC86I/video.html - you don't have to wait a month.
tks so much. any idea if we can connect Azure Open AI APIs in VF? tk u!
You definitely can
Excellent video, thank you very much! Can you tell me how to add an initial welcome message before the user enters their first question?
You need to do that on the Voiceflow side. So before the first "capture user reply" step, you need to add a text card with a welcome msg.
hello there, tnk u for the amazing video. i wanted to know if i'm supposed to have a current subcription with voiceflow in order to integrate my openAI assistants?
Yes you do need an active voiceflow account in order to do this. You can try the free sandbox but I’m not sure if it offers everything you need in order to set this up.
"Hi, thank you very much for the video and the template!! One question, how could I make it so that during interactions with the chat, it can remember past questions?"
With this assistants API setup it does remember past questions :)
Great video! Super simple low code Assistant api call! But i have a question, do you think this parameter OpenAi-Beta and value assistants=v1 in the api call header section will change? and so we'll have to change manually all the agents we built until then?
Appreciate that! The header will change once it is out of BETA. The assistants will most likely stay the same, would be very surprised if you have to manually change anything.
Make sense, i'm trying this template but the workflow always stop in the "check run" stemp beacuse it saying path not connected but all 3 conditions are connected instead, any chance you can help? Thanks!
@@OmniFusionAI
I think i got it, the issue only occur when retrieval using gpt3.5 turbo, with gpt4turbo it always work, maybe with 3.5 turbo the responses are slower and api stop to work, i think i'll have to wait beta to finish to have full capabilities
My only issue with Assistant APIs is that it is in beta and it takes time due to multiple statuses before it gets completed. How do I deal with it? Pls help!!
Yes, it's not quite ready for production in our experience. Rate limits and time delays are a big factor.
Thank you for sharing 😊👍
You’re welcome! 🤝
Great video! I have a question though. Is there a way to make it more conversational in ways of the openAI assistant remembering what the previous question was. Now, it looks like it only checks the latest question 'last_utterance'
The assistant will have a thread ID where it will have the conversation history saved!
@@OmniFusionAI Alright great! I'll look into that
thanks for this video ! I'm currently using the template but I have a problem when I send {last-utterance} to the POST API. When the variable is too long, it doesn't work :
API call error - {"status":400,"data":{"error":{"message":"Invalid body: failed to parse JSON value. Please check the value to ensure it is valid JSON. (Common errors include trailing commas, missing closing brackets, missing quotation marks, etc.)","type":"invalid_request_error","param":null,"code":"invalid
{"alt-svc":"h3=\":443\"; ma=86400","cf-cache-status":"DYNAMIC","cf-ray":"8a4bc5634a725797-IAD","connection":"close","content-length":"311","conte...
anyone with this issue ?
Are you sure it's because of length? I think it doesn't have valid JSON. You can work with ChatGPT to write JavaScript code that removes anything that would break the JSON, such as linebreaks for example.
Can you make a video about "How to add custom actions and API calls to your Custom GPTS on OpenAI"
I've added it to the list, but we are focused on API Assistants and all their applications more than GPTs currently. Subscribe and turn on notis so you find out when we do get around to making that video :)
@@OmniFusionAI subscribed and clicked the bell icon. Will follow your content ✨️
Is that a good way to reduce tokens usage?
With this solution you will be charged by openai directly, and not use any voiceflow tokens at all.
That's amazing! thanks!!!@@OmniFusionAI
Hi. I have a doubt. I imported the voiceflow project you gave us. I created a customGPT using OPENAI's playground, replaced all API keys and assistant IDs in the voiceflow project. Here's the problem I countered
The chatbot is unable to create pdfs, ppts using code interpreter
Hey! I can't help with that currently since we haven't worked with those specific functions and actions yet. Once we get going on that, I'll make a video.
Hello I hope you're fine? The API is at version 2, does this template still work? THANKS
Should still work but haven't tested since we don't use it actively.
@@OmniFusionAI Thank you can you update it later please? I read the documentation I think you just need to change v1 to V2. I can't test it at the moment, that's why I'm asking you.
do u have a VF template for Azure open ai ? tk u
We do not currently
I get the following error, can you please help? 401 error with the message "Incorrect API key provided."
Did you check if the correct API key was provided? 😅
@@OmniFusionAI yes I did, my key is correct.having the same issue with other voiceflows
@@ivossos"I could resolve it with the following format: Bearer APIKEY (bearer begins with uppercase and the apikey without quotes)."
@@germanlujan OMG! thank you for this! I've been having issues ever since I used Assistants v2, I couldn't make my voiceflow run for weeks and I was stuck then I tried your suggestion and my problem was solved! Thank you so much!
Template not working with ChatGpt 4o unfortunately. Getting this error. Anyone know a fix?
The operation failed with an error. [400] The requested model 'gpt-4o-2024-05-13' cannot be used with the Assistants API in v1. Follow the migration guide to upgrade to v2: / docs / assistants / migration
You’ll need to use the assistants api v2 as described in the error message you got
@@OmniFusionAI I am using V2. It turned out to be a very easy fix. Anyone having the same problem, all you have to do is go into every template node and change "assistants=v1" to "assistants=v2" under "OpenAI-Beta". These manual references caused it to fail even though I was using V2 in the Assistant itself.
Great!