Hey Thunkers!
We’ve just introduced a new integration with OpenAI’s ChatGPT model. It’s very easy to use and add into your own apps! Take a look at our latest blog post below and let us know what you think! We’d love it if you could share some of the apps you create using ChatGPT.
I’ve created a demo app here that you are welcome to remix and play around with.
We also have a new set of docs on how to use Chat GPT.
Sorry if this question has already been made, but i can’t genuinly find the Open AI component, even tho i’m following these steps:
Navigate to your project’s Blocks tab.
Scroll to the bottom of the blocks panel on the left side to access the Advanced section.
Click the expand chevron to show the Advanced components.
Click the plus icon next to Open AI Services (Beta).
Your “Open AI Services (Beta)” component will appear under the Open AI Services (Beta) drawer. Click the gear icon next to the component’s name to access its properties dialog.
Any chance to get a screenshot from someone?
Thank you in advance for your time.
Here you go @maurizio.polverini89
I feel stupid, i searched for it but still can’t find it. Is it just for D&D ?
Yes, currently only available on Drag and Drop projects. Perhaps try a hard reset of x.thunkable.com if they are not showing up in your DnD projects.
Very cool! I’ve been working on my own app and am just about ready to share it. Interesting to now have a block that does a lot of what I figured out through trial and error.
What sources are you using for text generation and image generation? In my app, I’m using text-davinci-003 and DALL-E (not DALL-E 2), both from OpenAI. From what I’ve read, text-davinci-003 is the newest version available from their API but it is not as current as ChatGPT.
Also, I did not enter an API key. I simply opened the project you shared and it worked for me. Is it possible the API key is already stored in the demo?
Can you explain what these blocks do?
The demo you posted prefixes responses with formatting characters. Sometimes a “(”, sometimes “_matches”, sometimes “.jpg” etc.). Here’s an example:
This is cool
this is awesome , when remixed will not work , might be because the API , so can you please share the screen shot of API configuration and what has been used on Open_AI_Service for text and for image
@ubay.salihlzj Likely the API as I did not share my own personal secret API key. You can check our docs here about this.
Thank you very much for sharing this , and can you please elaborate further what we should use for Property and Value
i saw in the document mentioned ,
Property: Name of the property you are assigning a value to in your API request
Value: Value of the property you are specifying in your API request
so please can you share what has been used for Property and Value
@ubai.salih You would not need to add anything into either of those two fields to get started with our OpenAI integration.
and what we should use if we need to build it from scratch ? as you can see the description is not so clear for me , so please if you can elaborate further how to integrate with OpenAI it will be really great , because using only API key will not work and it seems there are more setting
it’s nice to know that we can connect to openAI via a thunkable block designed for that purpose.
first i asked “Tell me, in a poem, why python rocks” and it came back with this:
it seemed to have problems in handling newlines (\n) though,
so i split the returned response by hand:
Python’s great for all your coding needs
It’s easy to learn and a great tool to please
It’s open source, and it’s ever so flexible
And it’s got a huge library to explore and collect
Python’s great for automation and data science
It’s so versatile, it can do almost anything
It’s got a great community, full of helpful folks
Ready to help you when your code chokes
Python’s the language of choice for many
It’s got the power to solve any coding worry
It’s easy to use and understand
The perfect language for any tech-savvy man
Python rocks, it’s clear for us to see
It’s the language of choice for you and me!
actually i made an earlier request with this prompt:
write a poem describing why i should learn python.
and it came back with this (which is not bad) - i also had to split by hand.
Python is a language that’s worth learning,
Its power and flexibility make it appealing.
You can create things that are quite amazing,
And the learning process is not too taxing.
It provides a base for quite a few things,
From simple scripts to web development wings.
It’s easy to learn and quite a lot of fun,
And you can apply it to many a project, one and one.
The applications it has can be seen,
From automation to data science and machine learning.
So don’t hesitate, take the time,
To learn Python and you won’t regret your climb.
so it’s not ChatGPT but it tries!
then i tried the image generation with this prompt:
robot in grass skirt , green hills in background, style stained glass
i was surprised by the response - it may have missed the grass skirt (i should have worded it differently) but overall, the image is impressive!
then i tried the prompt “don quixote dancing with snow white in the style of picasso” and voila
there must be a way to interpret those newlines in the returned response of the first example for text generation. (when i was connecting to GPT-3 - directly - in the summer, the response came back with newlines correctly interpreted - perhaps in this format, via the control block, there is some setting i need to set).
all in all, it’s impressive - here are my blocks:
here;s the link to the project for remixing (you have to supplu an api_key)
https://x.thunkable.com/copy/c548528e5420ab6ce006d7e14e3c8f49
sign up for an API key with OpenAI and click on the link in @conroy33’s post above.
Please add to the original style interface (StP?). I will never move to DnD, it lacks: tools, features and clarity as well as the ability to “stretch” containers like one would on a website or fill a space like with CSS. It’s trying to be Webflow, without the functionality…
I’ve read the DnD feedback post and I have seen some good ideas, but to not bring new features to the original UI (though one could argue ChatGPT is useful for what - exactly?), seems like professional users are being forced to use a consumer interface (DnD)…
Is there a list of upcoming features that we can vote on?
What about integrations with 3rd party services like Make.com and n8n.io? Why is external data integration still just Airtable when there are literally hundreds of alternatives, many of which are better, or can be self-hosted.
I also find it odd that there is still no ‘search’ feature for any of the databases or tables. I know we can build one, but that seems like a feature that should have been around from the start, especially as a lot of Thunkable’s early promotion centered around “use a google sheet to create an app”, fine, but they can’t search it? Pretty useless to low-coders.
Hi team,
This is super cool. I can’t wait to start making things with the OpenAI integration.
I would like be to able to split the lines of text e.g. for a poem. How could I do this?
Hi ioannis
Thanks for your reply. I don’t think it will work because the length of the line will be different each time?
I’ll keep trying. Thanks 
