Categorize data with Airtable AI

Applying an AI field template

To quickly get started with categorization, you can create an AI field and choose a related template:

1. General > Categorization

2. Product > Feedback categorization

3. UX Research > Sentiment analysis

These templates make it very simple to configure an AI column. With General categorization, all you have to do is choose the field that contains your content and enter a comma-separated list of categories.

Content to be categorized: Hotel review

Copy and paste a comma-separated list of categories:

Safety, Cleanliness, Check-in, Food & Beverage, Price, Convenience, Room size, Noise level

Optionally, you can provide a few examples and provide additional instructions or guidance to the AI. Providing just 1-3 examples may significantly improve the quality of the categorization.

Adjusting AI field settings

You can enable the “Generate automatically” setting to have categorization automatically run when your field data changes.

By default, Airtable will select the best AI model, which may yield the best results but costs more credits per run and is slower. If you choose to not use AI models that optimize for the best results, then your text will populate more quickly

It’s recommended to use 0 for Randomness when categorizing, as you don’t necessarily want more “creative” answers.

Using AI field with custom prompts

These templates are a good place to start, and if you ever want to do something more advanced, you can move onto a custom prompt to build a more flexible AI query.

You can convert a templated prompt to a custom one by clicking the “Change to a text box” button:

This will allow you to see the prompt that was created based on your configuration in the template wizard.

Examples of things this enables you to do:

• Changing the AI persona to be specific to your use case

• Using multiple input fields

• Providing descriptions of the categories in addition to just their names

• Giving feedback examples that are more specific to your categories

• Changing the instructions, such as asking the AI to pick multiple categories instead of just one

You might use this edited custom prompt for tagging a hotel review with any of the relevant tags, not just one:

Using AI in Automations

If the AI field is not flexible enough for you, Automations provides a more advanced way of performing categorization. 

In Automations, you can create an AI action which makes a query to an AI model and gets back a response. You can then use that response in any of the ways you use Automations today—write an email, create a doc, post to Slack, write to a field in a base. In addition, you can input that response back to another AI action, creating the ability to chain AI calls together.

As an example of categorization, you can quickly create an automation to:

• Run when new feedback comes in via a form.

• Take the feedback content and a custom prompt that is configured in Automations and send it to AI to identify the feedback sentiment.

• If the sentiment is Negative, then post it to a Slack #feedback channel. If it’s positive, it posts to the #customer-love channel.

Try it now

Airtable AI makes it easy to categorize and label the data in your Airtable base. You can use a template to get started quickly, write a custom prompt for additional power, and use an automation for cases where a field might not fit.

Get started with Airtable AI today.