Learning Lindy

Looping

Learn how to run tasks in parallel with our Looping action

Looping allows you to efficiently execute tasks from a list of inputs, similar to a ‘for’ loop in programming. Instead of executing items one at a time, loops allow you to execute multiple items in parallel. It can unlock a plethora of powerful use cases, including:

A research agent researching many items in parallel.
An outbound phone agent conducting multiple calls at once and summarizing them all.
A lead generator researching and scoring 100s of leads at once.
An assistant summarizing all your meeting recordings of the past week.

Let's begin!

Approach

Here is a high-level approach for building a looping workflow:

Identify a list of items to loop through.
Define your loop output. What should the loop pass to the subsequent steps? For example, do you want a summary of the key findings or a specific value made available?
Define the steps within the loop. What should the loop do with each item?
Define what should happen after the loop.

Key Concepts and Terminology

Loop cycle - the task execution for a specific item.
Output - the result of the loop. The output of each loop is the only information from inside the loop that is passed to outside of the loop, so when configuring the output, think of what information you want to be available outside of the loop. If you do not specify an output, the output will just be a summary of the loop cycle.
Max Cycles - the maximum number of loop cycles that will be run for this Lindy. This is to help prevent infinite loops

Example: Research Assistant

Let’s build a research assistant that loops through each item to research, researches each item, and summarizes the key findings across all research items.

This research assistant can be customized to fit any research-related use case, including lead enrichment, candidate research, and deep research. Check out how in the Customize the Research Assistant section.

Going back to our high-level approach:

We need to identify a list of items to research. If you’re building a lead enrichment agent, your list of research items can be your lead list. If you’re building a recruiting agent, your list of items can be potential candidates. If you’d like to do deep research on a topic, your list can be different research angles around the theme of your research.

For this example, let’s research travel locations. We will supply the loop with a list of these items from Google Sheets.

We need to define the loop output. I want the loop to output a summary of the key findings for every location.
We need to define what we want to happen to each item in the loop. For each location, I want to search the web to find the average round trip cost from San Francisco, the average cost of accommodations and food for a week, and the top things to do. I then want these findings updated in the original spreadsheet.
We need to define what we want to happen when the loop ends. I want to send an email to myself summarizing key findings.

Let’s build this Research Assistant:

Setting up the loop

First, add a Message Received trigger so we can pass the items through to the loop.

Next, add a loop by selecting Enter loop in the next step menu.

You will need to specify the the items to loop through for the loop cycles, and the desired output of each loop cycle.

For items to loop through, you can provide a list manually or use values from previous steps. You can also use AI to define the inputs by putting this field in Prompt AI mode.

Our items to loop through will be the items from the ‘Things to research’ spreadsheet that we passed through the message received trigger earlier. Set the mode to Prompt AI, and enter “Research items from the previous step”.

For the output, I want a summary of the key findings for every location. Set the step to Prompt AI mode to define this output.

Now that you have your loop and output configured, it's time to set up the inside of the loop.

Building the loop

Inside the loop, I want to search the web to find the average round trip cost from San Francisco, the average cost of accommodations and food for a week, and the top things to do for each item. I then want these findings updated in the original spreadsheet.

Add a search action by clicking the + icon inside the loop. We have a plethora of search actions you can choose from: Perplexity, Search the Web, Google Search, etc.

Configure the action you chose:

To update the original spreadsheet with the findings for each location, add an Update Row action inside the loop.

Configure the Update Row action by adding the item output reference inside the Lookup Value field. The action will use the item that was passed through the loop to identify the row to update.

We want to update the ‘Findings’ column in the sheet with the findings from Perplexity. To do this, configure the action to pass in the response output reference from Perplexity to the ‘Findings’ column.

Your loop is all set up!

After the loop

Moving on, let’s configure what happens after the loop. I want to send an email to myself summarizing key findings. Add a Send Email action and pass in the results output reference from the loop to the body field of the Send Email action.

Your final Lindy should look like this:

Run your loop

Let’s run it. Turn it on or save it by clicking the button in the top right corner, and transition to the tasks view. Copy and paste your Research Items.

Watch all your tasks run at the same time:

You can also click into each loop cycle to see the separate task runs.

Watch it work:

‍

Spreadsheet output:

‍

Email output:

Customize the research assistant

The research assistant is a base that can be customized to fit a variety of use cases, including lead enrichment, candidate research, and deep research. Heres how:

Lead enrichment, candidate research, or any other people research: pass a list of people Items to loop through field. Utilize our People Data Labs actions, or our Linkedin scrapers within the loop to research your leads.

Deep research: Use our think action to brainstorm different research angles around the theme given by the user. Utilize a search action, such as Search Perplexity, Google Search, or Search the Web within the loop to research the research angle.

Guides

Building a Slackbot: 101

Tips For Reducing Costs