Skip to main content

​
What is Airtable?

Airtable Node

​
Node in Stack AI

Airtable is a cloud-based software solution that blends the features of spreadsheets and databases to offer a versatile platform for organizing, managing, and collaborating on content. Users can customize fields to include text, numbers, drop-down lists, attachments, and even links to other tables. This combination provides the flexibility of spreadsheets with the structure and relational capabilities of databases, allowing for varied applications ranging from content calendars and CRM systems to project management and inventory tracking.

​
How to use it?

Airtable in Stack AI

​
Connections to an Airtable node

To utilize the Airtable node, you must establish connections to both its input and output nodes.
  • Input: This node necessitates an Airtable formula. You can either specify the Airtable formula in an Input node, or use an LLM to transtale your request into an Airtable formula (see the template below)
  • Output: The node produces a JSON object containing data from the designated Airtable table. For viewing results, connect the output to an Output node. If you’re working with the LLM, ensure you’re aware of the context window. For extensive results, consider linking to a Vector database.
Additionally, you will need to link the node with an account. Click on the Connect to Airtable button and link it to your account. Once done this, you will be able to specify some required parameters.
  • Personal access token: your personal ID to identify yourself to Airtable.
  • Base: an Airtable “base” is essentially a database within the Airtable platform. Each base is made up of a collection of tables, similar to sheets in a spreadsheet or tables in a traditional database.
  • Table: this is the table object that you’d like to connect to.
  • Search Type: this is the mechanism over which to load data from airtable: - Formula: Runs an airtable formula to get the answer. Ideal for questions that involve looking for precise data matches “Give the emails of users older than 21 years”. - Semantic: Runs a semantic search over the rows of the table. Ideal for questions that look for facts in specific rows “List users with names similar to Andrew”.
Airtable in Stack AI

​
Airtable node connection parameters

​
Example of usage

A practical application of Airtable involves integrating it with an AI assistant designed to answer user queries based on context data stored in Airtable.
Airtable Template

​
Answering questions from users using Airtable data

To construct this workflow, employ an LLM to address the users’ inquiries. Incorporate an Input node to funnel the users’ submissions, and link an Airtable node to a dedicated account. Furthermore, a secondary LLM facilitates the conversion of user requests into Airtable formulas. The prompts for this supplementary LLM are as follows:

​
System prompt (LLM 0) 

You convert user questions into Airtable formulas. You only respond with airtable formulas that can be used as filters.

Example:

User: hello
Assistant: FIND("fox", "quick brown fox")

User: Is the word Bernardo in the name?
Assistant: SEARCH("Bernardo", {{name}}) > 0


User: What is the age of Bernardo?
Assistant: {{NAME}} = 'Bernardo'

CRITICAL: Your responses always are valid formulas.

​
Prompt (LLM 0) 

My Airtable has the following columns:

| Column | Type |
| < Name of Column 1> | <Type of Column 1> |
| < Name of Column 2> | <Type of Column 2> |

Answer the following question: {in-0}
For the other LLM (the one answering the user’s questions), the prompts used in this example are the following:

​
System prompt (LLM 1) 

You are a helpful AI assistant. You will be given data from an Airtable database
to answer questions. Be conversational with the user.

​
Prompt (LLM 1) 

Airtable formula used:

---

## {llm-0}

Response from Airtable:

---

## {airtable-0}

User message: {in-0}