Connect Airtable to OpenAI API without using 3rd party tools

More than two weeks and not a single comment or kudo. What does this data tell us?

1. It's possible the Airtable community is not ready for AI.
2. It's possible this message has gone unnoticed; perhaps this is part of the reason.
3. Is it possible that AI through script is scary; a bridge too far for most users?
4. You would think with all the noise about Enterprise sales and adoption, a few techy builders would be all over this.

Does anyone care a guess as to why AI is so important everywhere except here?

OpenAI's ChatGPT captures all data almost perfectly. But you have to let ChatGPT learn the most recent data first, then you can start working.

> OpenAI's ChatGPT captures all data almost perfectly.

No, it doesn't. You shouldn't make statements that are false. ChatGPT is aloof to your data and unable to learn unless you spend a non-trivial effort to make it do so. @Bikash has demonstrated a way to put a copilot in the cockpit with your app. ChatGPT doesn't even know there's a plane that needs to be flown.

> But you have to let ChatGPT learn the most recent data first, then you can start working.

In all generative AI solutions, the most recent data is often biased, and a fraction of the story that must to be shared for inferences to be valid.

This topic is about exactly the opposite of what you are attempting to sell. I suggest you start another topic with your insights and the value you can bring to the Airtable community. Raise your bar to include thoughtful ideas and why your ideas make sense.

@Bill_French I for one would love to use the script to populate my database with descriptions of images. But it appears to have been retired due to changes in the scripting tool, which means that "the script is failing to work, sometimes".

Is this it for that bit of code or will something nice happen to it ? 

No comment until I see actual code.

Sorry for the confusion, you weren't the one offering the script, it was @Bikash, the OP.
As I undesrstood it, there was meant to be a script available (against payment of a nominal sum) on :

I use Airtable script to communicate with at least a half-dozen different generative AI models. No issues.

Mind sharing how you do it? In times of GPT-3 I had a script working well. Only issue was the 30 second limitation that Airtable imposes for a script to finish. Have to re-write the whole thing to work with current models and coding is not my strong suite. So any inspiration for this would be appreciated.

>>> In times of GPT-3 I had a script working well.

This is the upside of using an LLM that no one cares about any more. 😉 It is performant because no one is using it.

>>> Only issue was the 30 second limitation that Airtable imposes for a script to finish.

Long-running inferences (or any long or complicated process) require you to separate Airtable script from the process. Use Airtable script only to kick off a process. Do not wait for that process to finish. Allow the process (in Replit or whatever) to run on its own timeframe and use the Airtable API to update target fields, etc.

Interesting. I have heard about Replit but worked with it. Is there a resource you can point me to explaining how you use it in conjunction with Airtable, ideally with example use cases?

Ha ha! No - that sounds like a project. 😉 Almost 100% of the stuff I build is guarded because clients don't want to see their efforts published.

Broadly, I use NodeJS with the Airtable API to build call-back systems that listen for commands, and record IDs, and respond accordingly by updating the target records/fields.

En fournissant des explications détaillées et en aidant à la compréhension de concepts complexes, Chat GPT Gratuit facilite l'apprentissage et l'enseignement dans divers domaines, rendant l'éducation plus accessible et interactive sur https://chatgptgratuit.net/

One way around this is Data Fetcher Airtable ChatGPT integration. You can make API requests to OpenAI that last up to 3 minutes, and there is a 30 minute limit for the total execution (e.g. updating the Airtable base).