Translate text with Airtable AI

What is AI-powered language translation?

AI-powered language translation is the use of artificial intelligence algorithms and machine learning techniques to automatically translate text from one language to another. The AI model relies on accurate inputs to produce context-driven translations. Over time, more and more input helps the model deliver more precise translations. 

Implementing AI translation in Airtable

Method #1 - Start with a template

The easiest way to get started with translation using the AI field is to use our pre-built translation template. 

To access the template, navigate to General > Translation. You’ll be asked to input a field with content, and a language to translate into.

The custom instructions box can be used to add specific context for your translation. For example, you can add a note about the formality of the language or specific terms to use or avoid. 

Take a look at the comparison box below to see the difference between translations with vs. without context. In this example, we’re asking the AI to translate button text on a shopping website from English to Spanish. The first column notes the term in English, the second column shows the translation with no context added, and the third column shows the translation with the context “translation for buttons on a shopping website”.

Method #2 - Use a custom prompt

If you’re interested in building a more flexible AI query, you can convert your templated query from method #1 into a custom one by clicking the “change to a text box” button.

The original prompt is very simple:

Here are a few examples of how you might use a custom prompt to meet a more specific need than the generic translation template: 

• Provide the name of the company (“You are a copywriter at Airtable…”) to get the model to provide translations that match those of the company.

• Provide a glossary of terms that you want used or excluded.

• Create a field for storing language and populate the language via the user’s address, for example. Then use the field reference in the prompt in order to translate the message into a different language for each user.

• Use AI fields to translate input from a user into English (for example) for analysis, then craft and translate a response to that user. This requires 4 fields:

  • Field 1: Translate user’s input into English

  • Field 2: Detect the user’s language

  • Field 3: Craft a response to the user in English (so it can be reviewed)

  • Field 4: Translate the response into the user’s language

• Translating a document into multiple languages and incorporating guidelines docs

  • If you’re translating a document into multiple languages and have different sets of guidelines, you can set up linked record field rollups to pull in the correct guidelines.

    • Create a linked record table where the language is the primary key and the guidelines are in another field in the table.

    • Create a record in the original table for each translation you want to do.

      • Link each record to the correct language in the linked record table, then use a lookup field to pull the guidelines for each record from the linked record table.

      • In the AI field, pull in the copy, the language, and the language guidelines.

• Translating hierarchy/ taxonomy

  • If you’re interested in translating product taxonomy such as the filters and categories on a shopping website, it may be useful to provide the context of the taxonomy above the current item.

    • For example, if you were trying to categorize “Trivets”, it might be useful to know that those are part of a category called “Cookware.”

    • You could set up a translation system using 3 tables with different levels of linked record hierarchy, so that Kitchen & Dining > Cookware > Trivets. 

    • You could then use lookups to pull the data from the category above (and potentially below) into the record to provide context for its translation.

Method #3 - Use Automations

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

Here are two examples of ways you could use AI in Automations:

Example #1: Use Automations to translate a new piece of customer feedback into English and post it to Slack using the following logic: 

Examples #2: Take input for a blog post via a form, use AI to generate the blog post, translate the blog post into Spanish, write it to a Google doc, and email the user a link to the Google Doc using the following logic:

Try it now

Airtable’s new AI features make it easy to incorporate language translation into your day-to-day workflows. For more straightforward use cases, you can rely on the pre-built translation template to translate text from one language to another, with the ability to add relevant context. For more advanced use cases, you can rely on custom prompts, and even incorporate AI into your logic flows within Automations.

With Airtable, you can harness AI-powered language translations to communicate with a global audience, scale your work, and connect with new people & cultures around the world.

Get started today with Airtable AI.