ฝัง
- เผยแพร่เมื่อ 20 ก.ค. 2024
- 🔗👩💻 Embed Your Custom GPT to ANY Website for FREE!! 🚀🎯
🎟️ COUPON : 0PUCWSYC
(First 200 users 1 month free)
🔋 OAI-Widget:
www.oai-widget.com/
🚀Custom AI Solutions:
www.vrsen.ai/
🤝Need help with your next AI project?
calendly.com/vrsen/ai-project...
🐦Twitter:
/ __vrsen__
About: Learn how to easily and affordably embed an OpenAI Assistant on your website. We show you how to bypass expensive services and set up a cost-effective Custom GPT widget, enhancing your site's efficiency and sales. This step-by-step tutorial covers customization and using OpenAI's Assistants API without coding, making it ideal for businesses and individuals keen on creating their own Custom GPTs.
Intro - 0:00
Setting Up Your First Widget - 1:34
Conclusion 7:37 - วิทยาศาสตร์และเทคโนโลยี
![[TH] VCT Pacific Stage 2 - Playoffs Lower Bracket Final // PRX vs DRX](http://i.ytimg.com/vi/5HwO1bQyAx4/mqdefault.jpg)
Always learning something new from your videos! Great job!🔥
I've been looking for a solution like this for months. Other companies charge too much for too little it's ridiculous. Great work!
Subscribed, set up in 5 minutes and working perfectly. Well done and thanks for the free month. This is simply a winner.
Great work boss. 6k already. Much deserved. I feel your growth curve finally starting to get steeper. Let's go!!!!
Thanks man! Let’s get it!
@@vrsen BTW I managed to run the Agency-Swarm colab and I liked it a lot, then I installed A-S from github, but there is no main or app, do I need to build it myself with the snippets in the Readme filer?
I don't understand the workflow, how do I get started, what do I run?
Just found your channel brother. Really enjoying and appreciating the content you are putting out! Keep up the great work. Would love to hire you sometime.
Always fast and informative content with VRSEN. Subscribed. Would love to see a future video focused on LLM chaining in a "safer" language like Rust.
Your videos are awesome! I hope you keep brining on more knowledge.
Love it! Congrats and well done!!!
Thanks Lucas!
I'm so happy I found this video. You made it so simple. Thank you!
Great stuff!! Congrats and thank you! I just tested it and it works very well!
Thank you so much for this video. Super helpful and well-presented. Your time and effort is most appreciated!
Great content. Thank you for sharing!
Super cool! Is there a way to see the transcripts from interactions people have with the widget once it's embedded?
Always amazing information. Thanks
Amazing tool! There's must be an analytics feature built in since my clients require track the number of interactions, the session length and other. Other than that, super helpful. Fantastic work Arsen! Keep it up!
Thanks man, saw your email. We can add analysts, no problem. I will reach out in a bit for more details
@@vrsen Better make a video of how you do it
Thanks@@vrsen , yes, analytics is very important, great video!!
This is my new favourite channel
Just signed up. Love what you’ve been laying down.
Thanks man, don't forget the coupon! 0PUCWSYC
Great content as always :) Nice chatbot solution. Does the chatbot offer the option to add something like conversation starters (suggested messages) presented as buttons with question inside? I think it would very helpful as some users don't know how to start the chat or what to ask and the other chat solutions have this feature.
Awesome video. With great power comes great responsibility.
Hey thanks a lot for building this was searching for something like this from the past month. Just to reiterate in the free tier we can ask unlimited questions right? and we will only be charged by Openai API key right? There is no cap on the questions by OAI?
Do you have to always create an assistant from scratch within OAI platform or can I use the assistants that I have previously created withing the openAI assistant platform? Because in the dropdown I cannot see any of my previously created assistants within OpenAI. Thank you!
❤ super content que TH-cam me présente cette vidéo ! Merci VRSEN ! Abonné illico presto !
I am no code guy... mine is working!!! Thank you
You make such great content! My one request would be that you add streaming for a better user experience.
It’s not supported by assistants yet, unfortunately, but thanks for feedback!
I like the OAI-widget and consider going premium, but the website is super slow. WIll that soon be fixed?
Is it possible to use another LLM Api? Like llama 2 from a personal server?
Is it safe to share your OpenAI key with the OAI-Widget? OpenAI writes (as you can see at the video at 3:01) "Do not share your API key with others, or expose it in the browser or other client-side code. In order to protect the security of your account, OpenAI may also automatically disable any API key that we've found has leaked publicly."
Thank you so much for this!
🙏
How do you train this GPT on all your youtube videos though? Makes sense for training on a website but what if you have 100s of videos?
Great video and tool. I am using it now. Would love to see a feature to use external databases.
You can easily do this with actions!
Good Sharing！ Thanks
I wonder if they are going to add a toggle to the Assistant API to turn off its ability to send multiple responses per run? I'm not really a coder but I've been playing around with the API in colab. Printing the latest message is straight forward but figuring out is there are multiple assistant messages and outputting them in the correct order is beyond my feeble skills. I could prompt it in the system message to never produce follow-up messages but what's the fun in that?
It's been nice not having to deal with managing the context length like before.
I think it's only multiple tools at once, not responses
@@vrsen The assistant API has the ability to add two or more separate responses to the thread in the same run. I actually had a detailed conversation with it about the feature. Later I had it create a prompt for me that often gets a multi part answer. I just tested it again and it sent three back. From my testing code interpreter might be what's allowing it to happen so it needs to be enabled.
Prompt: "Hello AI, I’m interested in understanding how you handle multi-part messages in a conversation. Could you please demonstrate this by first describing your process and then providing a separate follow-up with additional insights or tips?" Sometimes I'm unlucky and it won't work. Sometimes you can ask for a list of things and add something like, "...then do an additional follow-up message with details on one of them" to the end.
I also had it create a prompt to prevent them but I've not tested it thoroughly.
Oh, wow, I’ll test it out thanks. My guess is that this is due to code interpreter, but not sure
Hi
I have not been able to add a new assistant in oai_widget
it says "error adding new assistant, please try again"
how do i solve this?
Great tool!🥳 Are you planning to add streaming response capabilities in the future?
Of course! Hopefully openai releases this soon
1:07 you got me there😆
how to extract a bot created from gpt store for e.g whatsapp or something?
Thanks! Already built one. The only thing I can't quite figure out is how to embed my chatbot in an and not as a widget. It would be great if you could share how to do that 🙏
Hey, not possible yet but, web apps coming up soon. It will also have chat history when embedded as a web app
Cool! How does this compare to other builders like Botpress and Voiceflow though - any advantages?
1. You have full control over your OpenAI Assistant and your prompts.
2. You pay only for what you use. We do not overcharge you for OpenAI API.
3. You can connect any API just like on ChatGPT.
I have noticed the responce time is quite slow for the chatbot compared to other options I have tried. Are there plans to speed up the responce time for the chatbot?
Thanks for feedback. Primarily this is due to GPT outputting large chunks of text. The longer the response - the bigger the latency. Make sure to add at the end of your instruction “never output messages longer than 3-5 sentences each.” This will help. Additionally, would you like us to add gpt-35 model? It is quite stupid, but responses would be much faster
@@vrsen Please add the ability to connect artificial intelligence Gemini
Ok so it can take quite a while before it loads the text so whynot for the sake of the user add the text "Typing" to where it is loading so the user knows that the actual responce is coming.. because I just did a test now and it took 20 seconds and many users will just close the chat and move on .. people have short attention spans.@@vrsen
How do i contact you i want to know the security of this. If to be used for clients.
Does it work with vision in a custom GPT?
it only shows "gpt-4-1106-preview" not even gpt4-turbo that was released in April, not to mention gpt-o.
Also, it is not about openai "assistant" API
So, this widget allows not only to answer customers to their questions, but also to save the customer's question or the customer in chat mode can fill out a form for purchase, contact the support team?
Yes, exactly! With custom actions the possibilities are virtually limitless!
@@vrsen That's awesome! The process of customer acquisition and keeping customers will be taken to the next level with this tool
@mykhailoschuka yes, our front end dev did a great job
Hi there. I bought a $5 widget, but am having trouble configuring it. I found Voiceflow easy to configure with lots of training. Do you have step by step detailed training info to support this widget implementation? When I loaded the widget on a site, none of the configuration came through and the bot did not recognise my knowledge base. I would also like more info on how to actually create actions . Is this something I need to research elsewhere? Thanks
i created a video and an article which I would in the description on using oai-widget to build AI Assistants if that helps
Hey :) The code has expired. Any chance getting a new one? 🙏
Great work! Did I get it right that with the free version I can't upload files and not enter any instructions?
You can upload files and set instructions, for sure. You just can’t remove the watermark or add actions.
There is also a limit of 3 files per assistant on free, but if you upload files through openai yourself, it’s unlimited as well.
I have to say it's a great product for someone with very limited AI knowledge like myself.
I have signed up for the pro version but I am curious about something.
If I create my assistant on the OpenAI website, add the actions and files there (basically created on OpenAI and not In house), then will the actions work on the free forever plan as well?
No, unfortunately there is no way to support actions for assistants created through OpenAI. However, if you create your assistant through OpenAI, there is no limit to how many files you can upload. (Currently, free only support 3 files, at most 5MB each if you create it on oai-widget.com)
Thanks@@vrsen Also the default model is GPT4. Can we change the model to 3.5 while going to GPT 3.5 dashboard?
You can change when creating on openai. We will add gpt3 to our app this week
@@vrsen amazing. Looking forward
This is done btw
How do you store your TH-cam scripts in the Assistant API and can it send a source link back to the video itself when referencing where it made up it's answer from?
Yes, export all TH-cam transcripts in a PDF or Markdown file. Include links to the corresponding videos in the document. Instruct the model to reference these links when using information from these transcripts.
thanks@@vrsen. What tool are you using for that process please?
I already had all my transcripts in notion. For editing you will probably have to do this manually or write a python script
What is the video you mentioned about creating endpoints to firebase?
th-cam.com/video/voJuGHTfO_g/w-d-xo.html
I want this, but only as an to put the chatbot into another locations, not a widget. Is this possible?
What other locations?
How to setup to send emails?
But is it possible to charge clients for using my custom made gpt for niche? can i make some payment plan etc?
Like a GPT store? No, but you can create a GPT and sell it to multiple clients. There is no limit to how many clients you can sell it
I am trying to upload a 26kb json file and i get an error saying you cannot upload more than 100kb to open ai.
I started on a free plan, then a paid plan and I get the same result, so am stuck.
Yea, this is openai issue. I am not sure why it currently has this error
Hey mate this looks great, but the price on the website is $15 per month. Have you put the price up?
No, it’s for custom GPTs. Widgets are still 5
🔋www.oai-widget.com/ Let me know your feedback❤👇
How does one upgrade to the paid plan mate? Not seeing the option to within the platform. Thanks
@@tonyblack2141 it's near the toggle to remove the watermark when you edit your widget
@@vrsen Got it. Thanks
Cant make public without pro? Frustrating to follow all the way through build something and then find I cant toggle to public without paying for pro? So I cant embed in a website for free, which was the whole title and point of the youtube. Am I correct? The custom GPT "Public" toggle is greyed out.
You can embed widget for free, but custom gpts is our new feature. It is a lot more expensive to run.
My feedback for pro version: Not otpimized for seo (page speed slow) Not optimized for mobile, Great for assitant direct from ai platform (instructions) and great for thread on platform for lead but don't work fine on site :(
Check your widget, the default message there is typo: "OpanAI".
Lmao, thanks
@@vrsen As a feedback: when running on colab there are 2 things I noticed:
1. If you create too many agents and your Company Manifest is too long you will you overload the API, so maybe you want to cap the manifest, proportionally to the model used?
2. But also when you create the Agents their prompt is the Company Manifest only? Wouldn't it be better it to be a concatenation of Manifest + Instructions?
Or the script on Github is different? That's why I wanted to try it, I can't find an entry point no main.py nor app.py, I don't understand what I need to do to get started? Do I need to create the agents programmatically first?
@@truehighs7845 Agents will combine both manifest and instructions. Don't want to cap it, to ensure that users have full control. If anything, OpenAI will return an error. To test it you need to run run_demo method from your agency. See the collab notebook
Please help. It doesnt work on IPhone. Why? Your service is great, but how i can resolve it
Can you please submit contact form on our website?
How is this different than rest of the no-code, low-code tools like voiceflow, stack ai etc. ?
They offer free package aswell.
You have full control over your AI, and you pay only for what you use. Unlike their free packages, ours offers you a lot more capabilities via OpenAI's Assistants API with UNLIMITED messages.
ok got it. would be cool to see a breakdown in a table format or else to see the differences pros and cons. so I can share with other communities and then they can make decision@@vrsen
Landing is not our top priority right now. I explain this briefly in the video, and our own chatbot also has this information. Feel free to share these for now. Thanks in advance
Good video but I wish you went more into detail about how to put the gpt on a website.
I tried to sign up and got an error "Network Request Failed."
I could log in using Safari but nothing happens when I click on API Keys.
how to connect custom trained GPT with oai-widget?
You can simply select it from drop-down on OpenAI tab
Hey Arsenii
I like your product but I had some issues when using the widget. I sent you a message with my problems from the feedback form from your website a few days ago but I have not gotten any message back.
Is there a different way to contact you?
Best regards
Hani
Круто! Спасибо большое! Как раз искал оптимальный вариант где делать ИИ-чатботов
I love this
♥
@@vrsen I just signed up and looking for how to upgrade now.
@@matthew_brown cool! Let me know your feedback!
Ok but what about your assistants instructor code 🤩 can’t wait to be able to add my own python functions and get that working with an agency
Yes! If we receive enough requests for this feature we will add it. For now, you can do exactly that by following this tutorial: th-cam.com/video/voJuGHTfO_g/w-d-xo.html
Of course, make such instructions, there is a shortage of such videos on the Internet, everyone is busy with gossip, every day there is a lot of artificial intelligence, I’m starting to get confused in this, a lot of water. And you provide us with cream
How do I upgrade to pro? Unable to find link on site to upgrade?
It’s near remove watermark when you edit your widget
Thanks bro have gone Pr. Please also is image size for assistant icon - keep saying error upload
@@vrsen
The website GOOD
Doesn’t Flow wise do this already?
Yes, if you want setup and pay for servers yourself, plus struggle through creating OpenAI Assistants with Langchain.
Can you use this on social media sites
Not yet. Which one should we add first?
@@vrsen Pinterest or you tube
Where can I get the API key free? Please tell me 😔
On platform.openai.com but without adding a billing method you can only use gpt-3.5
Bro, you have a Slavic accent, I don't know if you are Russian, Ukrainian, or Belorussian, but your accent is very Slavic, but that's not the point. You are cool all over the internet, everyone is posting videos of how to connect artificial intelligence to python and use it on your computer, 99% are filming about it, it's so annoying. And you're the only one who's figured out what's the use of it just being on a computer. It should be put on the web and people should use it, what is the use of these tutorials if they are limited and no one shows how to put it on the web commercially.
I hope he is not Russian nazi
You make a great point about too few putting the agents onto the internet
please add the ability to send voice messages and the ability to connect a bot telegram.
You got it man! We can add voice pretty fast. Telegram might take longer however
agent swarm is a cool tool! thanks for doing this!
I think it would be cool to expand the features on your site and add:
1. incoming and outgoing voice,
2. make not only a widget, but the possibility to get a separate mini site with password login for my friends.
3. save old conversations in separate tabs (archive without continuing the conversation).@@vrsen
@@vrsen Hello! We are really looking forward to these new features :)
Great tool, great video. I started to use the widget. If the info is retrieved from the knowledge base, at the end of an answer I get e.g.【8†source】. I assume it shouldn't be there...
This is GPT citing sources. We will try to filter this, but for now just tell the model not to cite documents in its responses
This is fixed now
I subscribed 2x and yet 😩. I dont know what went wrong that I cant make it work.
What’s the problem?
Not simple at all (and I'm an engineer).
Subscribed to Pro, this is exactly what I was looking for I have been using a different company but their development is to slow. I have an idea for a specific function for a chatbot I think would help many not only users but entreprenuers. I will contact you.
Yes let us know your feedback please through our widget or contact form
What company did you use
Great work ! But not quite there ! The embed script has many errors - logger.ts:115 [2023-12-15T03:42:50.719Z] @firebase/app-check: FirebaseError: AppCheck: ReCAPTCHA error. (appCheck/recaptcha-error).
at pL.getToken (providers.ts:169:27)
logger.ts:115 [2023-12-15T03:42:50.722Z] @firebase/app-check: FirebaseError: AppCheck: ReCAPTCHA error. (appCheck/recaptcha-error).
at pL.getToken (providers.ts:169:27)
at async ZM (internal-api.ts:156:5)
at async Promise.all (index 1)
Uncaught (in promise) FirebaseError: Missing or insufficient permissions.
Thanks, looking into it. Could you send your website link please?
Its not live ,it a simple html ;
Chat Component Example
document.addEventListener('DOMContentLoaded', function() {
// Check if the chat container exists
var chatContainer = document.getElementById('chat-container');
// If the chat container doesn't exist, create it
if (!chatContainer) {
chatContainer = document.createElement('div');
chatContainer.id = 'chat-container';
document.body.appendChild(chatContainer);
}
// Initialize the Chat component
if (window.ChatComponent) {
ChatComponent.init('lkzzhfY5UgazV9tKXCCb','#chat-container');
} else {
console.error('ChatComponent is not available');
}
});
You can delete comment when done.
@@vrsen
@adautomation1585 could you please provide more details on how to reproduce?
Easy I just opened the html file and saw the errors in console.
@@vrsen
Does it work at your end? Could it be a security issue?
I tried the widget but get an error. Wrote to the chatbot and waiting for the response. Can you please help me?