Help Center
How can we help? 👋

What is Levity and who is it for?

A simple explanation of who we are, what we do, and who we’re for.

Levity is a no-code tool for automating text, image, and document tasks that previously required human decision-making.

Our AI technology uses supervised learning (humans helping the AI to learn) to automate repetitive tasks involving unclassified data (messy/scattered data such as freeform text and images).

How does Levity actually work?

A good analogy for explaining how Levity works with Blocks and Flows is to compare it to a conveyor belt assembly line in a factory.

AI Blocks represent the individual machines or workstations along the assembly line, each designed to perform a specific task. These tasks could be attaching a part, painting, or packaging. In the context of Levity, AI Blocks are trained to classify text, images, or PDFs based on the categories you define.

Notion image

Flows represent the conveyor belt that moves items through the assembly line, connecting the various workstations. Your Flows ensure a seamless transition from one AI Block (workstation) to another, coordinating the overall process. In Levity, Flows are the in-built automation templates that connect AI Blocks to various tools and services, such as Gmail or Google Sheets.

Just as a factory's assembly line can be customized with different machines and workstations to create a unique product, Levity allows users to create custom AI Blocks and integrate them into tailored Flows, automating their specific tasks and workflows.

What is the overall process to set up Levity?

It’s very easy! Our process is just a few steps:

  1. Start with the Flow Builder: Go to the Flow Builder on the Getting Started page and select your trigger app, then choose your desired trigger action (e.g. new email). This will initiate the process of creating a custom AI Block for your specific use case.
  1. Prepare the data: Gather a collection of data containing examples of what you want Levity to learn. The data can be a folder of images, a digital file of PDFs, or a spreadsheet of text examples. You can also import data from a selection of tools. You’ll need at least 2 labels (the things you want Levity to detect for you), and 20 pieces of data per label.
  1. Import the data: Use our smart uploader to get your example data into Levity. Assign labels to each piece based on what you want to identify (e.g. label the positive review “Positive”, and the negative review “Negative”). These labels will help train the AI Block to recognize patterns associated with your data labels.
  1. Train your AI Block: Using Levity's interface, one-click train the AI Block to classify the labels identified in your training dataset. The on-screen wizards will guide you through the process.
  1. Test the AI Block: Evaluate the AI Block's performance by testing it with additional, previously unseen instances of your examples. Make any necessary additions and retrain to improve the accuracy of the AI Block.
  1. Complete the Flow: After the AI Block is performing at an acceptable level of accuracy, finish setting up your Flow by defining the final action (e.g. logging the prediction in Salesforce).
  1. All done: With the AI Block integrated into your Flow, Levity will automatically classify new instances of your data based on the labels you've defined, streamlining the process and connecting it with your existing tools and systems. Voila!
Still too much work? Clone out our Templates and do it in just a few clicks!

Who uses Levity?

Because our solution is so simple, Levity is simply for everyone. From individuals who want to use Levity for private uses to big corporations that do not want to focus on building up their own AI team, we can solve problems for all kinds of users. We work across industries with companies of all sizes.

Is Levity for you?

We made this part super easy - try before you buy! We have a free trial, so you don’t need to spend a penny to save time and money with Levity.

If you have questions you'd like answered, give us a shout at

Want to know more about what our users have built?

Here are some examples:

Did this answer your question?