10 Low-Code Prompts for Airtable, Bubble, and Retool to Ship Faster

Low-code and no-code platforms have changed how we build software, but the paradox is that they still require a good deal of technical thinking. You need to design data models, write formulas, connect APIs, and reason about logic. That’s where AI prompts come in. Instead of staring at a blank schema or a confusing workflow, you can ask an LLM for a starting point, then iterate. Below are ten prompts I actually use in my own workflow, split across Airtable, Bubble, and Retool. Each one is battle-tested and comes with a concrete example so you can copy, paste, and adapt.

Airtable Prompts

1. Generate a Table Schema from a Description

This is my go-to prompt whenever I’m starting a new base. Instead of manually creating fields, you ask the AI to design a schema, including field types and sample records.

Prompt:

Act as an Airtable database architect. Design a basis for tracking freelance projects. Include tables for Clients, Projects, and Payments. For each table, list all fields with their types (text, number, date, single select, etc.), and add 3 sample rows of realistic data.

Example output:

This prompt works because it forces the AI to think about relationships and field types, which you can then refine in Airtable. Always review the output—AI can miss edge cases like archived clients or multi-currency budgets.

2. Write a Formula Field for Date Calculations

Airtable formulas can be tricky, especially when you need to handle null dates or business days. I use this prompt to generate and explain a formula.

Prompt:

Write an Airtable formula that calculates the number of working days between two date fields (Start Date and End Date), excluding weekends. If either date is empty, return "N/A". Explain what each part does.

Example output:

IF(
  OR(IS_BLANK({Start Date}), IS_BLANK({End Date})),
  "N/A",
  (
    DATETIME_DIFF({End Date}, {Start Date}, 'days')
    - 2 * INT((WEEKDAY({Start Date}) + DATETIME_DIFF({End Date}, {Start Date}, 'days')) / 7)
  )
)

The AI will explain that INT((WEEKDAY(...)) accounts for weekend boundaries. In practice, you might also want to handle holidays—you can extend the prompt by adding a custom holiday table.

3. Automate Notifications Based on Status Changes

Airtable Automations need a trigger and a set of steps. This prompt helps you sketch out the entire automation before you build it in the UI.

Prompt:

Create a step-by-step Airtable automation: when a record in the Projects table has its Status set to "Complete", send an email to the linked Client’s email address with a summary of the project (name, budget, completion date). Use pseudocode for the steps.

Example output:

1. Trigger: When record updated
2. Condition: If Status is "Complete"
3. Look up Client email from the linked Clients table
4. Send email: Subject "Project [Employee Name] is complete!"
5. Body: "Hi [Client Name], the project [Project Name] is finished. Final budget: [Budget]. Completed on [Today]."

You can then plug these steps into Airtable’s automation builder. It’s much faster than reading the docs and writing the logic from scratch.

Bubble Prompts

4. Design a Database Schema for a Marketplace App

Bubble’s data types and fields are easy to misdesign. I use this prompt to get a clean relational model before building UI.

Prompt:

Act as a Bubble database expert. Design the data types for a two-sided marketplace that connects dog walkers and dog owners. Include fields, field types, and privacy rules for each type.

Example output:

• User (extending Bubble’s built-in User): name, bio, location, role (walker/owner), rating (number)
• Dog: owner (User), name, breed, energy level (option set)
• Booking: walker (User), owner (User), dog (Dog), date, status (option set), price
• Review: author (User), booking (Booking), rating, comment

Privacy rules: Users can only edit their own records, walkers can see bookings assigned to them, users can only view reviews for their own dogs. This gives you a starting point for Bubble’s point-and-click privacy settings.

5. Generate Workflow Logic for a Feature

Instead of reading Bubble’s workflow documentation, I ask for the exact steps in Bubble’s editor language.

Prompt:

Write a Bubble workflow for a simple sign-up: when a user creates an account and has selected "owner" as their role, create a new Dog record with placeholder data. Use Bubble's workflow action names (e.g., Sign the user up, Make changes to a thing).

Example output:

1. Trigger: When User signs up
2. Condition: Current User's role = "owner"
3. Action: Make changes to a thing (Create a new Dog)
4. Set fields: Dog's owner = Current User, Dog's name = "New Dog", Dog's breed = "Unknown"

You can paste this into Bubble’s workflow editor. The AI won’t click the buttons for you, but it gives you the exact sequence of actions and conditions.

6. Build a Repeating Group with a Query

Repeating Groups are the core of Bubble lists. This prompt helps you configure the data source and structure.

Prompt:

I have a Bubble app with a Booking data type. I want to show a page with today’s bookings for the logged-in walker. Write the steps to set up a Repeating Group, including the type of content, the data source, and the list of fields to display.

Example output:

• Type of content: Booking
• Data source: Do a search for Bookings, where Walker = Current User, and Date = Today
• Repeating group layout: Display the Dog’s name, time, and owner’s phone number in a row
• Add a button to mark the booking as "Completed"

This saves a ton of time when you’re building admin dashboards or customer-facing lists. You’ll still need to style it, but the logic is done.

Retool Prompts

7. Write a SQL Query for an Internal Tool Table

Retool works well with SQL databases. This prompt generates a query that you can paste directly into the query editor.

Prompt:

Write a PostgreSQL query for a Retool table that shows all orders from the last 30 days, with the customer name and status. Filter to only unshipped orders, and order by order date descending.

Example output:

SELECT o.id, c.name AS customer_name, o.total, o.status, o.order_date
FROM orders o
JOIN customers c ON o.customer_id = c.id
WHERE o.order_date > NOW() - INTERVAL '30 days'
  AND o.status != 'shipped'
ORDER BY o.order_date DESC;

Retool’s table component will automatically map the columns once you set the query as the data source. I often copy the result and tweak the filters using Retool’s UI dropdowns.

8. Create a Transformer Function for an API Response

Retool Transformer functions are JavaScript snippets that reshape data. This prompt is perfect when you need to convert a REST API response into a table-friendly format.

Prompt:

Retool transformer: I have an API response that returns a list of users with name, email, and nested address object containing city. Write JavaScript that flattens this into an array of objects with columns: name, email, city. Handle the case where address may be null.

Example output:

const rows = data.map(u => ({
  name: u.name,
  email: u.email,
  city: u.address?.city || "Unknown"
}));
return rows;

This is a common pattern in Retool, especially when connecting to legacy APIs. The optional chaining (?.) is supported in Retool’s JavaScript environment, so you get clean, defensive code.

9. Validate Form Inputs Without a Plugin

Retool forms let you set validation, but complex rules are easier to prototype as JavaScript. This prompt gives you a reusable validation snippet.

Prompt:

In Retool, I have a form with fields: email, age, and country. Write JavaScript that returns a validation object (e.g., {email: "Invalid email", age: "Must be 18+"}) for a submit button. Use regex for email and simple condition for age.

Example output:

const validation = {};
if (!/^\S+@\S+\.\S+$/.test(email)) validation.email = "Invalid email";
if (age < 18) validation.age = "Must be 18+";
if (country === "") validation.country = "Country is required";
return validation;

Attach this as the Approve event handler on your submit button. It’s faster than using Retool’s built-in validators when you need custom regex or business rules.

10. Scaffold a Retool Workflow with Temporary Tokens

Retool Workflows are great for orchestration. This prompt helps you plan a workflow that calls an API with a temporary token and then writes the result to a database.

Prompt:

Write a JavaScript step in Retool Workflow that: 1) fetches a CSV file from a URL, 2) parses it, 3) maps rows to an update query, and 4) returns a summary count of rows processed.

Example output:

const csv = await fetch("https://example.com/data.csv").then(r => r.text());
const rows = csv.split("\n").slice(1).map(line => line.split(","));
// Assume update row by row for each entry
const count = rows.length;
return { count: count };

Then you can chain that into a Retool Query step that upserts the rows into your database. Workflows are where Retool shines, and starting with a prompt like this cuts the boilerplate down to almost nothing.

Final Thoughts

These prompts are not magic—they still require your judgment. Always test generated formulas and queries against real data, and don’t blindly trust AI output for production logic. Instead, treat prompts as scaffolding: they get you 80% of the way, and you finish the remaining 20% with your domain knowledge.

The best part of this workflow is that you learn, too. When an AI generates a formula, take the time to understand what it does. Over time, you’ll need fewer prompts because you’ll internalize the patterns. Until then, keep this list handy. Copy a prompt, adapt it to your situation, and ship your next tool in hours, not weeks.