There's a bug right now if you don't "train the assistant" when testing that leads to variables being reported as 0. Hit "train assistant" and this resolves itself :)
I'm triying to implement this too, did you just lowered the number of prompt before calling chatgpt? And is there a way to keep the record of previous questions or does it always opens a new one?
The response AI and set AI steps are now supported on Alexa! So no need to make a custom API call to GPT now. Just add your prompt into the response AI step and you'll be good to go!
Great video! Glad Voiceflow could be helpful here :) This is one of the usecases I was thinking of when we built the generate step - allows you to hit OpenAI directly from within Voiceflow - no API calls needed!
Thanks Braden. It is an honor. I did initially try with your A.I Generate step, but there was a few reasons I opted not to - a) It's output was only giving me around 115 words even when I had requested more in the prompt b) The credits felt like they would get used up quickly, in comparison to the new cost of using the API directly. Aside from that, really liking the platform. FYI, I have had a few users state there is an issue with variables returning zero, atm, I managed to replicate - I have reported this issue. Also, while you are here, is it possible to create skill for Google home in a similar way? (I have no idea about G-Home, but people seem rather interested. Thx
@@promptnewsAi I ran into this same "0" issue despite following your tutorial closely, though it sounds like you encountered it yourself later as well? I suspect it has something to do with the Alexa/Amazon/AWS not set up to handle the input correctly, but then I wonder why it worked for you in the video. Frustrating! Please do reply if you find a solution, I've been wanting ChatGPT on Alexa forever
Thank you very much! Since the ChatGPT arrived I wanted to do something similar. It is so interesting that Siri, Alexa, and other assistants become so old type (however many things can do better, then ChatGPT)
I am having troubles with stories that are 400 words long, if I try 40 that works. I think voiceflow does not wait long enough for the answer from openai? Any ideas ?
This is absolutely cool that this works! The fact that it's hardly anything for tokens and it's quite accessible to setup different programs and codes.
Thanks for this video! I was able to build off of it to create a more typical prompt/response conversation which was a lot of fun. Voiceflow is very cool, don't know how I've never heard of it
did you not face any issues in Voiceflow testing or Alexa developer console testing? I keep getting an error while Alexa testing that the Skill did not provide a valid response
NOTICE: ALL people getting JSON errors: 1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT. 2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai. Here is a copy of the corrected formatted json: { "model": "gpt-3.5-turbo", "messages": [{ "role": "user", "content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. " }], "temperature": 0.7 }
Could you integrate with Siri or have a way to get this working on mobile? Also as others have said, is it possible to get this to be more of a general use voice assistant? Cheers!
HI, something not working .... When i'm trying to send a request, for testing, i have that message : "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON" do you knwo how i can solve it ? Thanks
NOTICE: ALL people getting JSON errors: 1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT. 2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai. Here is a copy of the corrected formatted json: { "model": "gpt-3.5-turbo", "messages": [{ "role": "user", "content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. " }], "temperature": 0.7 }
OMG just commented on your other video because this documentation kept referencing the api options.. and I couldnt find them. Why did they hide it like that!? haha thank you as usual
Chat interface for GPT ChatBot by Lacora s.r.o. Rated: Mature 4 Free to Enable ”Alexa open chat interaction” ”Tell me what is the tallest building on earth.”
I am getting the error message: "An exception occurred while dispatching the request to the skill" in the test developer console. I am suspecting that is related to a timeout consuming the ChatGPT API. Did you have this problem?
I've been getting timeouts or hit-and-miss responses via Alexa. I believe that if the chatgpt doesn't respond within a specific time, the alexa times out and has a silent failure. Works fine using my assistant, but I get weird responses on the Alexa side.
I can't modify mine. I created an account in both Chat GPT and Voiceflow, but that was about it. Voiceflow won't even let me develop a flow, it wants me to put in a website to get started? I don't have any website? I think I copied an API address, but I do not know how to put it in. The menu's are completely different than what is depicted in this video. Any help would be appreciated.
Sorry for the late reply - I wanted to check my facts before replying. Unfortunaly, it seems that this is not going to be possible on Google (or not for long at least), as Google has announced it is retiring the ability for users to create conversational-esqe experiences in their Home platform - developers.google.com/assistant/ca-sunset - a thus programs that allowed you to create the, including Voiceflow, are also removing the ability.
@@promptnewsAi Ah, thank you very much for looking into this, Yes Google seems to be nuking their own system. We've lost features via IFTTT recently and I'm finding Google Assistant to be going down hill. I hope some clever person can build a speaker and put ChatGPT into it without the need of any other tool. I'd much rather speak to ChatGPT than Google now. :-)
Hi I got most times a for 99% an server time out with " response time out" - 408 server error. Sometimes it gives me a response. But when it gives a response what part do I need to fill in de Capture response? The whole response or partly? And witch part because I got stuck on those 2 issues. Please help me out! Thanks PS: I see this in my test request by headers: text/html; charset=utf-8 while i put in a json format etc?
I was trying to connect to chatgpt with Aws and it was a mess, gpt4 was taking up to 2 minutes to respond, I'm gonna try your approach, thanks very much!
Great content congrats! I'm starting to get familiar with this kind of things, can you tell me if it's possible to program that same chatgpt skill in a different language, i'd say alexa to respond in portuguese for example?
Good video, it was very helpful. I made it to the testing section, but how can I use it on my own Alexa without having to publish it on the skill store?
In the write up, I have added screenshots and pasted extaxtly what my API section said, you may use that to compare - promptmuse.com/step-by-step-guide-to-integrating-chatgpt-3-5-turbo-into-your-alexa/
Chat interface for GPT ChatBot by Lacora s.r.o. Rated: Mature 4 Free to Enable ”Alexa open chat interaction” ”Tell me what is the tallest building on earth.”
@PaulSonnneholznerS-wf2qo read the comment that was pinned here in the comment section. He claims to have fixed the issue by following his instructions, but I haven't personally tried his fix yet. Let me know if it helped.
@@PaulSonnneholznerS-wf2qo Click Run in the top right then before clicking the blue button to start the run, there's a rounded "page" for training directly above.
Chat interface for GPT ChatBot by Lacora s.r.o. Rated: Mature 4 Free to Enable ”Alexa open chat interaction” ”Tell me what is the tallest building on earth.”
I want to see you put 5 ChatGPT in a room and have them all start talking to each other, and then we just listen in for 10 hours to see what it's up too .... :-)
I actually did that by just copying and pasting texts of 2 chatgpt browsers. But after like a few responses it will get into a loop about whatever topic and it just keeps saying the same thing basically.
This is great but it seems a little small bore to hook Alexa into ChatGPT just to funnel it into such a narrow prompt. What's the best way to set it up for general questions? Like can you say "Alexa, ask ChatGPT [whatever]" and have [whatever] passed directly to the API for a response? Edit: after some tinkering it may be possible by starting with a "Listen" block and passing to the API for a response, but there's currently some sort of bug that interprets the user input as "0"
i've accomplished this by maintaining a variable {mesage_history} that accumulates the JSON value of the "messages" JSON object. The ChatGPT API expects you to upload every prompt and response so far with each new prompt (so it has context of the history). So, capture user_input, add {"role":"user", "content": "{user_input}"} to the message_history, POST {message_history} to the ChatGPT API, capture the chatgpt_response like seen in the video, add {"role":"assistant", "content":"{chatgpt_response"} to the {message_history} then speak the {chatgpt_response} and repeat. I also used conditional blocks to look for illegal characters on the input and the response and sanitized them using javascript
@@maskettaman1488 I like what you're doing here, but I lose you for a moment when you say "add {"role":"user", "content": "{user_input}"} to the message_history". I've tried doing this after the capture and before the API post with Set Variable but i get an unexpected token error. Do you perform the input add in the API block itself? If you wouldn't mind explaining just a little more, it would be greatly appreciated. Thanks!!
Hi, i am trying this and I get an error when I send request to test API, but it gives me error, says can not parse json, can you help? I wrote the same code as you did. Thanks
NOTICE: ALL people getting JSON errors: 1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT. 2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai. Here is a copy of the corrected formatted json: { "model": "gpt-3.5-turbo", "messages": [{ "role": "user", "content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. " }], "temperature": 0.7 }
Dude it worked in the test section and i tried checking distribute privately but i didn't get it please help me so that i can connect it to my echo device🙏
Assist a beginner please. I was able to successfully execute the tutorial on Voiceflow, but upon attempting to run it on Amazon, everything runs smoothly until the final execution where an error is returned indicating "The requested skill did not provide a valid response" what am i doing wrong ? thank you
NOTICE: ALL people getting JSON errors: 1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT. 2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai. Here is a copy of the corrected formatted json: { "model": "gpt-3.5-turbo", "messages": [{ "role": "user", "content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. " }], "temperature": 0.7 }
NOTICE: ALL people getting JSON errors: 1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT. 2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai. Here is a copy of the corrected formatted json: { "model": "gpt-3.5-turbo", "messages": [{ "role": "user", "content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. " }], "temperature": 0.7 }
Seems there were quite a few issues some people were having, which I why I have decided to create a new, improved and far simpler method.. hopefully coming soon!
I tried this, it didn't work properly (some of the error messages turn up in other comments here), and I used, I'm not kidding, GPT-4 for debugging. That helped to iron out the kinks (after a few iterations). I'm serious.
Chat interface for GPT ChatBot by Lacora s.r.o. Rated: Mature 4 Free to Enable ”Alexa open chat interaction” ”Tell me what is the tallest building on earth.”
You don’t go into any detail about the Alexa skill publishing. How do you publish it without making it publicly available or easily discoverable? Those tokens will go quick if someone figures out the right invocation prompt.
When you click publish as shown in video, it publishes it ONLY for your local use - I so when you see me using it on my Alexa, this is not accessible for public use - but instead tied only to my Alexa account. Making it public has its own set of hurdles and hoops you need to go through..
@Prompt News ok thanks a lot for that. I'm not sure of the steps to make it only usable with my own Alexa account. I clicked distribute, but based on the questions, it looked like I was going to launch it publicly, so I stopped there. How do you publish it only for your personal use? Am I missing something obvious? Thanks a lot for the great video. I have it working to answer any prompt I give and it's working great in the Amazon Developer testing environment.
Awesome video! I have tried to get it to work but I always get the "We could not parse the JSON body of your request." error. It also says in the header section of the test request 400 server error. I have copied your headers, body, parameters, and response exactly, but I still get the error. Thanks for your help!
Figured something out that solved this. My formatting in the Body section was slightly off somehow. I copy and pasted code again from Open AI's website and the format looked orange instead of black. Then it worked.
Hi, thanks for the tutorial. I have come to one issue, when I click send request to test it, its says "response timout" Headers Content-type. text/html; charset=utf-8 408 Server Error Any idea? thanks
Why isn't this a product? I want a website of this I can log into with my Amazon credentials and have it enable ChatGPT on my local Alexa's smart home control features.
Chat interface for GPT ChatBot by Lacora s.r.o. Rated: Mature 4 Free to Enable ”Alexa open chat interaction” ”Tell me what is the tallest building on earth.”
I'm lucky if I can figure out how to use my phone 😂 They need to just make chatgpt its own device like the dot. My alexa see.s like its getting dumber and when you compare it to chatgpt, there's a vast difference. Chatgpt is far superior. Make a chagpt dot-like device and I'll buy it!!!
Unfortunately, VoiceFlow is glitching so badly. It cannot understand simple phrases unless I scream them repeatedly. The voice to text makes it unusable. It also randomly goes off script, and I have no way of debugging this.
Chat interface for GPT ChatBot by Lacora s.r.o. Rated: Mature 4 Free to Enable ”Alexa open chat interaction” ”Tell me what is the tallest building on earth.”
Great content, but I was just about to subscribe and you didn't the whole please please please working hard here passive aggressive stuff... I'll watch a few more and probably subscribe anyway though
ChatGPT is quite helpful yet it lacks quite a few things and that is the ability to access the internet, show me photos, videos and everything I would find in google. While I do use it sometimes, not as much as the first few days. It is quite limited to help me in my daily life. It would be cool to have it attached to alexa, yet I still do not even have her hooked up in my office. I kinda want to get one of the echo shows instead of a dot. The one question? That I really have is how does the 18 dollars play in to ChatGPT? Do I have to use it up to keep using it? I have not even hit a number on it yet. I also do not know if I use it right anyways as it tends to forget and tends to not listen to what I ask it to do. It does make quite good use when using recipes yet it keeps messing it up when trying to plan out a dinner. Tends to spit out slightly diffrent thing all together. From what I gather gpt4 is suppose to be way better and I have signed up for it a long time ago.
There's a bug right now if you don't "train the assistant" when testing that leads to variables being reported as 0. Hit "train assistant" and this resolves itself :)
Thanks for coming back to feed this back to us, much appreciated. I have pinned the comment for all to see.
I am getting the following error: "Entity names cannot contain duplicate entries" every time that I train the assitant
whwere is the button : "train assistand" in voiceflow ?
@@AntonioJPisani i have the same problem
it still puts out 0, even after training...
This is awesome, thanks! I modified mine to just capture one voice prompt so I can ask ChatGPT anything I want, blows Alexa away :)
Can it be called like "Alexa, ask ChatGPT + user question" ?
I'm triying to implement this too, did you just lowered the number of prompt before calling chatgpt? And is there a way to keep the record of previous questions or does it always opens a new one?
Try this one: "Alexa, open Oracle of Light"
The response AI and set AI steps are now supported on Alexa! So no need to make a custom API call to GPT now. Just add your prompt into the response AI step and you'll be good to go!
how do we make it?
Great video! Glad Voiceflow could be helpful here :) This is one of the usecases I was thinking of when we built the generate step - allows you to hit OpenAI directly from within Voiceflow - no API calls needed!
Thanks Braden. It is an honor. I did initially try with your A.I Generate step, but there was a few reasons I opted not to - a) It's output was only giving me around 115 words even when I had requested more in the prompt b) The credits felt like they would get used up quickly, in comparison to the new cost of using the API directly. Aside from that, really liking the platform. FYI, I have had a few users state there is an issue with variables returning zero, atm, I managed to replicate - I have reported this issue. Also, while you are here, is it possible to create skill for Google home in a similar way? (I have no idea about G-Home, but people seem rather interested. Thx
@@promptnewsAi I ran into this same "0" issue despite following your tutorial closely, though it sounds like you encountered it yourself later as well? I suspect it has something to do with the Alexa/Amazon/AWS not set up to handle the input correctly, but then I wonder why it worked for you in the video. Frustrating! Please do reply if you find a solution, I've been wanting ChatGPT on Alexa forever
@@Jordan117 I have since contacted support, and they are aware of this current bug, and are in the process of fixing it.
@@promptnewsAi Great to hear, thanks! Will try again tomorrow
They have just written to say they have fixed that bug.!
VoiceFlow has discontinued integration with Alexa, and it is no possible to do it this way. Could you please share the Skill files?
Nice Tutorial, ⚠FYI⚠The publish to alexa feature has been deprecated. To connect your Voiceflow project to Alexa you can do so using the Dialog API.
Thank you very much!
Since the ChatGPT arrived I wanted to do something similar. It is so interesting that Siri, Alexa, and other assistants become so old type (however many things can do better, then ChatGPT)
Try this one: "Alexa, open Oracle of Light"
Thank you! That is really cool you figured this out...I don't know squat about this stuff but I was thinking that this would be neat!
Try this one: "Alexa, open Oracle of Light"
I am having troubles with stories that are 400 words long, if I try 40 that works. I think voiceflow does not wait long enough for the answer from openai? Any ideas ?
This is absolutely cool that this works! The fact that it's hardly anything for tokens and it's quite accessible to setup different programs and codes.
Try this one: "Alexa, open Oracle of Light"
Amazing! Your voice sounds more mature, but I'm impressed that an 8yr old came up with all of this. ;)
LOL🤣
Thanks for this video! I was able to build off of it to create a more typical prompt/response conversation which was a lot of fun. Voiceflow is very cool, don't know how I've never heard of it
did you not face any issues in Voiceflow testing or Alexa developer console testing? I keep getting an error while Alexa testing that the Skill did not provide a valid response
@@sathish316 No, didn't have any issues at all
@@sathish316 I'm getting the same. It works or it doesn't. I believe it is time-based as in it may buffer too much and then get a bogus response.
Try this one: "Alexa, open Oracle of Light"
Excellent video!❤
Thanks! It's amazing
Excellent Alex, that was amazing. Imagine the possibilities.
Try this one: "Alexa, open Oracle of Light"
Is there a 4.0 Omni?
NOTICE: ALL people getting JSON errors:
1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT.
2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai.
Here is a copy of the corrected formatted json:
{
"model": "gpt-3.5-turbo",
"messages": [{
"role": "user",
"content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. "
}],
"temperature": 0.7
}
Well done!
Try this one: "Alexa, open Oracle of Light"
Could you integrate with Siri or have a way to get this working on mobile? Also as others have said, is it possible to get this to be more of a general use voice assistant? Cheers!
th-cam.com/video/THeet9bbphw/w-d-xo.html
HI, something not working .... When i'm trying to send a request, for testing, i have that message : "We could not parse the JSON body of your request. (HINT: This likely means you aren't using your HTTP library correctly. The OpenAI API expects a JSON payload, but what was sent was not valid JSON" do you knwo how i can solve it ?
Thanks
NOTICE: ALL people getting JSON errors:
1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT.
2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai.
Here is a copy of the corrected formatted json:
{
"model": "gpt-3.5-turbo",
"messages": [{
"role": "user",
"content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. "
}],
"temperature": 0.7
}
Lmao I thought you said it was free????
Agreed. What happens after 3 months?
Agee 100% not at all free, grossly inexpensive and borderline free, but not free
Hi folks! Just an update, Voiceflow no longer supports a one-click integration with Alexa but you can do so with our Dialog API :)
You are greeeeeeeat man .. thanks for the inspiration.
Try this one: "Alexa, open Oracle of Light"
OMG just commented on your other video because this documentation kept referencing the api options.. and I couldnt find them. Why did they hide it like that!? haha thank you as usual
Try this one: "Alexa, open Oracle of Light"
Would it be possible to add some more code so you could ask it more details after it has finished telling the storry? Make it more interactive ?
Chat interface for GPT ChatBot
by Lacora s.r.o.
Rated: Mature
4
Free to Enable
”Alexa open chat interaction”
”Tell me what is the tallest building on earth.”
is it free? or do you have to pay something down the road upon usage?
You pay for the API, but it is very cheap.
Thta awesome. Thank you!
great video, thanx
my like button flashed when you said drop a like wtf is this new
Magic!
The tutorial is good . But the Voiceflow does not work now. Any answer?
I am getting the error message: "An exception occurred while dispatching the request to the skill" in the test developer console. I am suspecting that is related to a timeout consuming the ChatGPT API. Did you have this problem?
same here - any workarounds?
I've been getting timeouts or hit-and-miss responses via Alexa. I believe that if the chatgpt doesn't respond within a specific time, the alexa times out and has a silent failure. Works fine using my assistant, but I get weird responses on the Alexa side.
Yep, they seem to have added a timeout, so you must limit the amount of words :(
This is awesome.
I can't modify mine. I created an account in both Chat GPT and Voiceflow, but that was about it. Voiceflow won't even let me develop a flow, it wants me to put in a website to get started? I don't have any website? I think I copied an API address, but I do not know how to put it in. The menu's are completely different than what is depicted in this video. Any help would be appreciated.
Oh this is very good. Like others have asked is this possible on Google?
Sorry for the late reply - I wanted to check my facts before replying. Unfortunaly, it seems that this is not going to be possible on Google (or not for long at least), as Google has announced it is retiring the ability for users to create conversational-esqe experiences in their Home platform - developers.google.com/assistant/ca-sunset - a thus programs that allowed you to create the, including Voiceflow, are also removing the ability.
@@promptnewsAi Ah, thank you very much for looking into this, Yes Google seems to be nuking their own system. We've lost features via IFTTT recently and I'm finding Google Assistant to be going down hill. I hope some clever person can build a speaker and put ChatGPT into it without the need of any other tool. I'd much rather speak to ChatGPT than Google now. :-)
Hi I got most times a for 99% an server time out with " response time out" - 408 server error. Sometimes it gives me a response. But when it gives a response what part do I need to fill in de Capture response? The whole response or partly? And witch part because I got stuck on those 2 issues. Please help me out! Thanks
PS: I see this in my test request by headers: text/html; charset=utf-8 while i put in a json format etc?
@prompt News any answer for this issue?
same exact issue for me too. works perfectly in the voiceflow editor but as soon as it is published it times out
I was trying to connect to chatgpt with Aws and it was a mess, gpt4 was taking up to 2 minutes to respond, I'm gonna try your approach, thanks very much!
Try this one: "Alexa, open Oracle of Light"
can i know what the background music name?
Hiya, all Music was created by A.I using soundraw.io .
Does it work for any language? Like Brazilian Portuguese?
Funciona em 9 línguas eu acho, português BR tem
Awesome stuff!
Nice, I've wondered how to make a conversation bot.
Great content congrats!
I'm starting to get familiar with this kind of things, can you tell me if it's possible to program that same chatgpt skill in a different language, i'd say alexa to respond in portuguese for example?
Try this one: "Alexa, open Oracle of Light"
Good video, it was very helpful. I made it to the testing section, but how can I use it on my own Alexa without having to publish it on the skill store?
Try this one: "Alexa, open Oracle of Light"
I keep getting this error from the API: ""We could not parse the JSON body of your request." . Any advice on how to fix this?
In the write up, I have added screenshots and pasted extaxtly what my API section said, you may use that to compare -
promptmuse.com/step-by-step-guide-to-integrating-chatgpt-3-5-turbo-into-your-alexa/
Chat interface for GPT ChatBot
by Lacora s.r.o.
Rated: Mature
4
Free to Enable
”Alexa open chat interaction”
”Tell me what is the tallest building on earth.”
This isn't working at all for me. When I click to run the test, the response comes back with either a 0 or a 1 as the story inside of VoiceFlow.
me too
@PaulSonnneholznerS-wf2qo read the comment that was pinned here in the comment section. He claims to have fixed the issue by following his instructions, but I haven't personally tried his fix yet. Let me know if it helped.
@@aaroncatolico7550 it says , "train the assistant",. Where is that button in voiceflow ?
@@PaulSonnneholznerS-wf2qo Click Run in the top right then before clicking the blue button to start the run, there's a rounded "page" for training directly above.
I keep gettign an error "Path Not Connected" Any advice?
i cant link alexa for some reason
It does not work in my echo dot, do you know why? It perfectly works in the test section but not in my device
Chat interface for GPT ChatBot
by Lacora s.r.o.
Rated: Mature
4
Free to Enable
”Alexa open chat interaction”
”Tell me what is the tallest building on earth.”
@@psonnenh Thanks bro!
I want to see you put 5 ChatGPT in a room and have them all start talking to each other, and then we just listen in for 10 hours to see what it's up too .... :-)
I actually did that by just copying and pasting texts of 2 chatgpt browsers. But after like a few responses it will get into a loop about whatever topic and it just keeps saying the same thing basically.
Muchas gracias! Estaba buscando algo como esto justamente ayer. Mil gracias, abrazo
This is great but it seems a little small bore to hook Alexa into ChatGPT just to funnel it into such a narrow prompt. What's the best way to set it up for general questions? Like can you say "Alexa, ask ChatGPT [whatever]" and have [whatever] passed directly to the API for a response?
Edit: after some tinkering it may be possible by starting with a "Listen" block and passing to the API for a response, but there's currently some sort of bug that interprets the user input as "0"
Hey Jordan sorry about this, yes we've reported the bug - it's the "entire user reply" on the alexa channel. This is getting fixed right now
Having the same issue, the send request works fine but the tests don't
i've accomplished this by maintaining a variable {mesage_history} that accumulates the JSON value of the "messages" JSON object. The ChatGPT API expects you to upload every prompt and response so far with each new prompt (so it has context of the history). So, capture user_input, add {"role":"user", "content": "{user_input}"} to the message_history, POST {message_history} to the ChatGPT API, capture the chatgpt_response like seen in the video, add {"role":"assistant", "content":"{chatgpt_response"} to the {message_history} then speak the {chatgpt_response} and repeat. I also used conditional blocks to look for illegal characters on the input and the response and sanitized them using javascript
@@maskettaman1488 I like what you're doing here, but I lose you for a moment when you say "add {"role":"user", "content": "{user_input}"} to the message_history". I've tried doing this after the capture and before the API post with Set Variable but i get an unexpected token error. Do you perform the input add in the API block itself? If you wouldn't mind explaining just a little more, it would be greatly appreciated. Thanks!!
...
Hi, i am trying this and I get an error when I send request to test API, but it gives me error, says can not parse json, can you help?
I wrote the same code as you did.
Thanks
NOTICE: ALL people getting JSON errors:
1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT.
2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai.
Here is a copy of the corrected formatted json:
{
"model": "gpt-3.5-turbo",
"messages": [{
"role": "user",
"content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. "
}],
"temperature": 0.7
}
After sending the body raw request it shows 500 internal server error... Help!
What's the current state of this? Any chance we can get an updated/streamlined guide? I'm guessing this is a relatively popular video for the channel-
Dude it worked in the test section and i tried checking distribute privately but i didn't get it please help me so that i can connect it to my echo device🙏
How to control the Alexa Smart Home Devices with Voiceflow?
Or start a routine?
Try this one: "Alexa, open Oracle of Light"
Need to pay monthly for this service. :(
It doesn't work for me. All I get is the response: entity names cannot contain duplicate entries. wjat did I do wrong?
Assist a beginner please. I was able to successfully execute the tutorial on Voiceflow, but upon attempting to run it on Amazon, everything runs smoothly until the final execution where an error is returned indicating "The requested skill did not provide a valid response" what am i doing wrong ? thank you
did you manage to find the solution to this? Facing the same issue rn
Keeps me in a loop when I try it in voiceflow. Ergo, doesn't work. I tried changing the JSON-Body inout a lot. No success...
NOTICE: ALL people getting JSON errors:
1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT.
2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai.
Here is a copy of the corrected formatted json:
{
"model": "gpt-3.5-turbo",
"messages": [{
"role": "user",
"content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. "
}],
"temperature": 0.7
}
i try to log in and it doesnt bring up a menu. cant even get past the first step. pls help
Hi, tried to log in to where? You mean you can't access VoiceFlow?
@@promptnewsAi i tried loging into openai and my account doesnt let me access the settings :(
I'm doing something wrong, I get this error "message": "you must provide a model parameter" but in Body I put “model”: “gpt-3.5-turbo”
make sure you are doing a POST API call, not GET
it's not working anymore
NOTICE: ALL people getting JSON errors:
1. Parsing error, make sure you are not using typographic quotations (italic), the must be STRAIGHT.
2. The user aborted the request. - These seems to be a new cutoff implemented by Voiceflow to prevent long requests. Unfotunatly, the only way around this is to a) use a smaller amount to words. b) use their inbuilt ai.
Here is a copy of the corrected formatted json:
{
"model": "gpt-3.5-turbo",
"messages": [{
"role": "user",
"content": "Write a dramatised, story of 50 words, for a {Age} year old, about a person called {Name} on the theme of {Story} - make it appropriate for a person {Age} years old- and add a moral to the story. "
}],
"temperature": 0.7
}
Does anyone have a timeout issue?
Seems there were quite a few issues some people were having, which I why I have decided to create a new, improved and far simpler method.. hopefully coming soon!
I tried this, it didn't work properly (some of the error messages turn up in other comments here), and I used, I'm not kidding, GPT-4 for debugging. That helped to iron out the kinks (after a few iterations). I'm serious.
Chat interface for GPT ChatBot
by Lacora s.r.o.
Rated: Mature
4
Free to Enable
”Alexa open chat interaction”
”Tell me what is the tallest building on earth.”
You don’t go into any detail about the Alexa skill publishing. How do you publish it without making it publicly available or easily discoverable? Those tokens will go quick if someone figures out the right invocation prompt.
When you click publish as shown in video, it publishes it ONLY for your local use - I so when you see me using it on my Alexa, this is not accessible for public use - but instead tied only to my Alexa account. Making it public has its own set of hurdles and hoops you need to go through..
@Prompt News ok thanks a lot for that. I'm not sure of the steps to make it only usable with my own Alexa account. I clicked distribute, but based on the questions, it looked like I was going to launch it publicly, so I stopped there. How do you publish it only for your personal use? Am I missing something obvious?
Thanks a lot for the great video. I have it working to answer any prompt I give and it's working great in the Amazon Developer testing environment.
*Why am I watching this even though I don’t have one?*
Down the rabbit hole, you go. :)
17:47
Awesome video! I have tried to get it to work but I always get the "We could not parse the JSON body of your request." error. It also says in the header section of the test request 400 server error. I have copied your headers, body, parameters, and response exactly, but I still get the error. Thanks for your help!
I have the exact same problem
@@Liquid_DnB_music So do I
Have the same problem
Figured something out that solved this. My formatting in the Body section was slightly off somehow. I copy and pasted code again from Open AI's website and the format looked orange instead of black. Then it worked.
For Real Wow Future is now i guess
Hi, thanks for the tutorial. I have come to one issue, when I click send request to test it, its says "response timout"
Headers
Content-type. text/html; charset=utf-8
408 Server Error
Any idea?
thanks
Hey can someone pls help me everything is working fine as it should but not on my actual alexa
Edit: you have to press build on the website
Lambaaaada 😂
Why isn't this a product? I want a website of this I can log into with my Amazon credentials and have it enable ChatGPT on my local Alexa's smart home control features.
I think you have found your million dollar calling in life :)
Some of these app builders should accept Bat tokens, they're free
Chat interface for GPT ChatBot
by Lacora s.r.o.
Rated: Mature
4
Free to Enable
”Alexa open chat interaction”
”Tell me what is the tallest building on earth.”
I'm lucky if I can figure out how to use my phone 😂 They need to just make chatgpt its own device like the dot. My alexa see.s like its getting dumber and when you compare it to chatgpt, there's a vast difference. Chatgpt is far superior. Make a chagpt dot-like device and I'll buy it!!!
Ohh I think GPT just announced a massive price decrease for their APIs! Great news :)
This is really amazing though!!! Sooooo quick
Tokens wtf? Chatgpt is free ffs
great vid but i think its not really free as suggested by the title.
anyway, you did a great job with explaining everything! thanks for this video
It's free as in beer. Doesn't cost money to do.
Open the pod bay doors chat…
It's not free...
Google Nest, Google nest, Google nest
Unfortunately, VoiceFlow is glitching so badly. It cannot understand simple phrases unless I scream them repeatedly. The voice to text makes it unusable. It also randomly goes off script, and I have no way of debugging this.
Chat interface for GPT ChatBot
by Lacora s.r.o.
Rated: Mature
4
Free to Enable
”Alexa open chat interaction”
”Tell me what is the tallest building on earth.”
@@psonnenh it’s pretty bad, can’t do follow up questions. Not a conversation without that.
@@remusomega ChatGPT is excellent at responding to follow question. Eerily good.
@@maskettaman1488 Thank you, but I'm talking about the Alexa Skill not ChatGPT specifically.
@@remusomega Right... ChatGPT can hold a conversation which means if your skill is not doing that then it's a problem with your programming
wooooooow is this real??
Am I the only one here who doesn't understand shit!
ALEXA GPT 😬
Great content, but I was just about to subscribe and you didn't the whole please please please working hard here passive aggressive stuff... I'll watch a few more and probably subscribe anyway though
Nice video but the voice from magic sucks no one is going to use that crack robotic voice
@prompt news, would love to get in touch with you for how we can help you create more content like this :)
ChatGPT is quite helpful yet it lacks quite a few things and that is the ability to access the internet, show me photos, videos and everything I would find in google.
While I do use it sometimes, not as much as the first few days. It is quite limited to help me in my daily life.
It would be cool to have it attached to alexa, yet I still do not even have her hooked up in my office. I kinda want to get one of the echo shows instead of a dot.
The one question? That I really have is how does the 18 dollars play in to ChatGPT? Do I have to use it up to keep using it? I have not even hit a number on it yet.
I also do not know if I use it right anyways as it tends to forget and tends to not listen to what I ask it to do.
It does make quite good use when using recipes yet it keeps messing it up when trying to plan out a dinner. Tends to spit out slightly diffrent thing all together.
From what I gather gpt4 is suppose to be way better and I have signed up for it a long time ago.
in 17:48 minute about "capture response": how do I find the "Enter Key" ? In the video, its appear "response.choices[0].message.content"