Step 1: Reach the sign-up page

Go to the sign-up page. Sometimes, OpenAI requests solve a captcha to prove that you are human.

Step 2: Use a Google or a Microsoft account or enter your email

Use a Google or Microsoft account to sign up. Alternatively, you can use your own email, which might require additional verification steps.

Step 3: Add your name and company to your profile

Once you sign-up, you need to add your first name, last name, and organization name.

Step 4: Add a phone number

Add a phone number on the next page. OpenAI uses this to prevent abuse since users can try to sign up using bots.

Step 5: Enter the SMS code

You’ll get a text message with the verification code. Enter this code on the OpenAI registration page to complete your registration.

Step 6: Go to your profile > View API Keys

After the SMS verification, you are logged in. Then you need to go to the “View API Keys” page, on which you can manage your API access. To get there, click on your avatar, and click on “View API Keys.”

Step 7: Generate a new secret API key

The API access is done with secret keys. They are long text strings that work like a username and password at once. You send this key when you are running your API call, and OpenAI will assign that API call to your user.

For this reason, don’t share this key, and if it leaks, you can delete it.

Click on “Create new secret key” and store your API key in a safe place. You’ll see this key just once, and if you don’t copy it, you’ll need to delete it and generate a new one.

Step 8: Add API credits

Depending on when you sign up, you might get some free trial credits. If you didn’t get them you need to add your credit card on the “Billing” page.

You can check if you have any free API credits under “usage”.

Step 1: Create an Airtable base

Log in to your account and click on “create a base” on the left side menu.

Step 2: Select a workspace

Select a workspace where you want to add this base. You can create a new workspace if you want to.

Step 3: Rename the base and table

Using meaningful names is important not just to correctly identify your tables and bases, but also to refer to them in your code. For this reason, use names that you can quickly recall in your script. If you want, you can just copy the names from our example.

Step 4: Adjust the table columns

The column names are just as important. Also, changing the column format to allow long text might be useful since prompts and answers can get long.

If you want, copy the “Prompt” and “Answer” column names, since these are the names we use in the sample script.

Step 5: Open the extensions menu

Find the extensions menu on the right side and click on it to see which extensions you can add to your base.

Step 6: Click add an extension

If you don’t have any extensions, click on “Add an extension” to bring the extensions search tool.

Step 7: Search for the Scripting extension

Use the search box to find the Scripting extension and add it to your base.

Step 8: Replace the default code

The Scripting extension comes with a sample code snippet.

Replace that code with this script (make sure to use your API key):

//SETUP
//use your API Key here (keep the single quotes!)
const apiKey = 'your API Key';

//change the API URL here if you want to use a different API
const url = 'https://api.openai.com/v1/chat/completions';

//change the model here
const model = 'gpt-3.5-turbo';

// gets the prompts table
let table = base.getTable("Prompts");

// Prompt the user to pick a record
let record = await input.recordAsync('Select a record to use', table);

// if there's a valid record selected
if (record) {
	//further model settings and variables
	let prompt = record.getCellValue("Prompt"), temperature = 0.7, maxLeght = 256, topP = 1, frequencyPenalty = 0, presencePenalty = 0;

	// set up the API call variables
	let request = {
    	model: model,
    	messages: [{"role": "user", "content": prompt}],
    	temperature: temperature,
    	top_p: topP,
    	max_tokens: maxLeght,
    	frequency_penalty: frequencyPenalty,
    	presence_penalty: presencePenalty
	}

	// set up the connection variables, including API key
	const data = {
    	"Content-Type": "application/json",
    	"Accept": "application/json",
    	'Authorization': 'Bearer ' + apiKey,
	};

	// API connection
	let response = await fetch(url, {
    	method: 'POST',
    	body: JSON.stringify(request),
    	headers: data,
	});

	// convert the response to JSON
	let answer = await response.json();

	//extract just the OpenAI GPT answer
	answer = answer.choices[0].message.content;

	//display the answer
	output.text(`Answer: ` + answer);

	//let the user pick if they want to save it
	let shouldReplace = await input.buttonsAsync('Do you want to save this answer?', [
    	{label: 'Save', variant: 'danger'},
    	{label: 'Cancel'},
	]);

	if (shouldReplace === 'Save') {
    	// Update the record
    	await table.updateRecordAsync(record,{"Answer": answer})
	}

} else {
	output.text('No record was selected');
}

‍

Here are the main blocks of this code:

• Create the setup variables. Make sure to use your API key
• Create a field so users can select the prompt to answer, using the input.recordAsync('Select a record to use', table) line
• If there’s a record selected, get the prompt text from it, with let prompt = record.getCellValue("Prompt")
• Set up the API connection details, using the request and data variables
• Connect with the OpenAI API, with the fetch() function and using the POST method
• Extract the answer from the response. It’s inside answer.choices[0].message.content;
• Create a prompt to select if the answer should be saved, with await input.buttonsAsync('Do you want to save this answer?', […] )
• Save the record, if the user wants to: await table.updateRecordAsync(record,{"Answer": answer})

Step 9: Test the script

After you paste the code, click “Run.” You’ll see an option to select which record you want to use. Pick one of the records with a valid prompt. You should see it processing before it returns the OpenAI GPT answer.

Click “Save” If you want to keep it or cancel if you want to start again.

Step 10: Run the script from the main page

Close the scripting modal and click on “finish editing” to close the code editor. Now you can run the script to get answers from the main Airtable page. Just click “Run” and follow the same steps to select a record and save the answer.