The JSON around function calls is a bit more complex, but it is also documented on their website. To make it easier you could also try skipping the custom modules and just feed regular Make Open AI modules with text from the website.
Hi there, fantastic tutorial - thanks for sharing your knowledge! I'm curious if there's a way to implement a bulk trigger in the workflow to populate multiple Airtable rows with different URLs simultaneously? Could you guide me on how to approach this, and what kind of challenges might arise from that approach?
Yes I would use the JSON return option and ask the AI to return you a JSON list of URLs - you would need to give it a schema with a list and then iterate via that list. You can see here a bit similar example where I ask Open AI to return JSON czcams.com/video/oBilFNbCNrQ/video.html
@@BusinessAutomatedTutorials Thanks for the response. I may not have been clear earlier. What I'm looking for is a way to fill an Airtable base with text data from multiple URLs. I aim to automate sending HTTP requests to these URLs, extract the text content, and upload it to individual Airtable rows in a single action (triggered manually). My intention is to later integrate this data with ChatGPT, but my current priority is to efficiently pull raw text data in bulk (up to 1,000 links) using the fewest operations possible. I hope this better explains my objective
This is awesome, I'm definitely buying and trying this for a project I'm working on. Can I ask, if the OpenAI response includes markdown, how do we handle sending this from Make to Airtable so that it properly displays in the rich text field? Currently in my manual method where I copy and past OpenAI response into Airtable, Airtable is only displaying markdown as ** Text ** instead of just *Bold Text* for example. I have to add a space and then delete it for it to populate properly. Wondering if I will have the same issue with this method? 🤔
Airtable "Long Text" has an option to allow for "Rich Text" - when you toggle that it will allow the field to store markdown and display formatting. I think this would also work with copy paste.
hi, i´m building this an will try it out, but i´m struggleing with the webhook. Is in Airtable a kind of automation that will be triggered in the background or do i only have to edit the field Tigger button with my Webhook URL? This part seems to be missing
Hi in this case I am just using the Webhook URL to add to Airtable button. You press the button - this opens the URL ( and closes it automatically using the javascript). That triggers the webhook. It could be also done via automation , but here I thought I do something different.
@@russelpea this is simple windows.close() execution , as long as you have Javascript enabled this should run fine. Make sure window return the code. (The HTML page with code)
For some websites I'm getting this error: "Rate limit reached for gpt-4-0613 in organization myorganisation-number on tokens per min. Limit: 10000 / min". What would be the best way to handle this situation? This happens even with just one website/request. Thanks.
this means the context window is being exceeded. The GPT 4 has context window of 8,192 tokens - this sounds like the article and prompt exceed that. While it is difficult to cut input in Make based on exact token length - you might want to use the split() function to limit maximum length of input text.
Most likely not. Some basic auth approaches will take username and password as part of URL, but most modern logins are based on session cookies - you might be able to attach cookies to HTTP request in Make, but it might not work the way you intended or work long term.
Awesome Stuff!!! Love your workflows - keep them coming :) Just bought your flow but I ran into an issue... for some reason I don´t have a record ID I can insert in Airtable (Step 3). So there might be an issue between the webhook and airtable. Can you help me fix this issue?
Thanks! just bought. Sorry if this is a dumb question: do we have to activate the webhook in a 3rd party app? Or do you have a resource on how to set that up?
Hi Greg, this is awesome. I tried downloading the Make blueprint but I can't see any links for this unfortunately. I managed to get to see the Airtable base, but can't see the Make blueprint anywere. Any advice please?
Great stuff, and do you have a way and video to grab email and phone number ? The msot for me where U would be interested, will be to scan a website and know is technologie with 2 things that I would love if Facebook pixel is installed , Google Analytics and eventually if wordpress or not?? Balise is the most important for me as I selling Advertissing on Google ads and Facebook.. It's will be per fect tool. My website, their contact and audit then I can send the sequence emal appropriate
Thank you for your comment! I'm glad you enjoyed the video. As for grabbing email and phone numbers, you would have to instruct the GPT to get you those... results may vary. As for GA and FB pixel - these are fairly standards scripts - if you run regex through the page HTML you could see if there are those matches or not.
![Fine-tuning ChatGPT with OpenAI Tutorial - [Customize a model for your application in 12 Minutes]](/img/n.gif)
That custom search function you added is really interesting. I learn more every day
Glad you like it!
One of the more advanced make flows i've seen. thank you this will get me thinking.
Thank you so much for your kind words! I'm glad the video was able to inspire some new ideas for you!
best hack to add a second webhook to close the annoying opening of a window 🥳
Isn't it 😁
Amazing work! I feel so dumb watching this 😅 I have to study all of this
The JSON around function calls is a bit more complex, but it is also documented on their website. To make it easier you could also try skipping the custom modules and just feed regular Make Open AI modules with text from the website.
Indeed. You are so clever. Thank you. I have purchased most of your Gumroad content. Bravo!
Wow, thank you!
So awesome! It worked. Thank you so much!
Glad it helped!
Hi there, fantastic tutorial - thanks for sharing your knowledge! I'm curious if there's a way to implement a bulk trigger in the workflow to populate multiple Airtable rows with different URLs simultaneously? Could you guide me on how to approach this, and what kind of challenges might arise from that approach?
Yes I would use the JSON return option and ask the AI to return you a JSON list of URLs - you would need to give it a schema with a list and then iterate via that list. You can see here a bit similar example where I ask Open AI to return JSON czcams.com/video/oBilFNbCNrQ/video.html
@@BusinessAutomatedTutorials Thanks for the response. I may not have been clear earlier. What I'm looking for is a way to fill an Airtable base with text data from multiple URLs. I aim to automate sending HTTP requests to these URLs, extract the text content, and upload it to individual Airtable rows in a single action (triggered manually). My intention is to later integrate this data with ChatGPT, but my current priority is to efficiently pull raw text data in bulk (up to 1,000 links) using the fewest operations possible. I hope this better explains my objective
This is awesome, I'm definitely buying and trying this for a project I'm working on. Can I ask, if the OpenAI response includes markdown, how do we handle sending this from Make to Airtable so that it properly displays in the rich text field? Currently in my manual method where I copy and past OpenAI response into Airtable, Airtable is only displaying markdown as ** Text ** instead of just *Bold Text* for example. I have to add a space and then delete it for it to populate properly. Wondering if I will have the same issue with this method? 🤔
Airtable "Long Text" has an option to allow for "Rich Text" - when you toggle that it will allow the field to store markdown and display formatting. I think this would also work with copy paste.
hi, i´m building this an will try it out, but i´m struggleing with the webhook. Is in Airtable a kind of automation that will be triggered in the background or do i only have to edit the field Tigger button with my Webhook URL? This part seems to be missing
Hi in this case I am just using the Webhook URL to add to Airtable button. You press the button - this opens the URL ( and closes it automatically using the javascript). That triggers the webhook.
It could be also done via automation , but here I thought I do something different.
@@BusinessAutomatedTutorials Hi I can't have the javascript in the webhook response to close the window 🤔 Am I missing something?
@@russelpea this is simple windows.close() execution , as long as you have Javascript enabled this should run fine. Make sure window return the code. (The HTML page with code)
For some websites I'm getting this error: "Rate limit reached for gpt-4-0613 in organization myorganisation-number on tokens per min. Limit: 10000 / min". What would be the best way to handle this situation? This happens even with just one website/request. Thanks.
this means the context window is being exceeded. The GPT 4 has context window of 8,192 tokens - this sounds like the article and prompt exceed that. While it is difficult to cut input in Make based on exact token length - you might want to use the split() function to limit maximum length of input text.
Perfect, I'll try that, thanks a mill👍
Could I visit websites behind a login if I put in username and password in http post request first?
Most likely not. Some basic auth approaches will take username and password as part of URL, but most modern logins are based on session cookies - you might be able to attach cookies to HTTP request in Make, but it might not work the way you intended or work long term.
Awesome Stuff!!! Love your workflows - keep them coming :) Just bought your flow but I ran into an issue... for some reason I don´t have a record ID I can insert in Airtable (Step 3). So there might be an issue between the webhook and airtable. Can you help me fix this issue?
I have the same issue which resulted in "Account doesn't exist: Connection not found 'airtable:1300486'."
You need to replace every Airtable connection with your own for if to work. In every single module.
Make sure you added correctly webkok under the button in Airtable and added ?recordId= and after that record ID
Thanks! just bought. Sorry if this is a dumb question: do we have to activate the webhook in a 3rd party app? Or do you have a resource on how to set that up?
Disregard, got it all worked out. Thanks, this is completely amazing!
LOVE IT 😁😁 Have fun and let me know what you build with this!
Hi Greg, this is awesome. I tried downloading the Make blueprint but I can't see any links for this unfortunately. I managed to get to see the Airtable base, but can't see the Make blueprint anywere. Any advice please?
If you get the version with Make blueprint there will be a JSON file to download : businessautomated.gumroad.com/l/access-web-with-openai-gpt-api
@@BusinessAutomatedTutorials That makes sense, thank you :D
Great stuff, and do you have a way and video to grab email and phone number ? The msot for me where U would be interested, will be to scan a website and know is technologie with 2 things that I would love if Facebook pixel is installed , Google Analytics and eventually if wordpress or not?? Balise is the most important for me as I selling Advertissing on Google ads and Facebook.. It's will be per fect tool. My website, their contact and audit then I can send the sequence emal appropriate
Thank you for your comment! I'm glad you enjoyed the video. As for grabbing email and phone numbers, you would have to instruct the GPT to get you those... results may vary. As for GA and FB pixel - these are fairly standards scripts - if you run regex through the page HTML you could see if there are those matches or not.