<aside> 📜 TABLE OF CONTENTS
Product: Sola Version: V.2.1 Updated: 3/18/2025
In this section, we'll explore how to use Sola to build workflows.
This section will cover the steps for recording and importing your workflow to the platform.
The first step in creating any automation with Sola is to record yourself going through the workflow - we operate under a show, don’t tell system.
To do this, open our browser extension, navigate to the starting page of your workflow, select record, and proceed through your workflow as you usually would. Once you are finished recording, you can import that workflow to the dashboard!
<aside> 💡 Note that for recordings with authentication, you’ll need to record your credentials in the workflow – otherwise our remote browser will not be able to access the same pages you are accessing when executing. We have some automatic checks for this in place and will notify you if that processing fails.
</aside>
https://app.arcade.software/share/WSiM77xkk1KAaMWkAguk
Finish Recording
Once you are finished recording your workflow, you can select this button to save what you have recorded.
Scrape Values
If you would like to capture text from a website to be used as an output or an input later on in your workflow, you can use the scrape feature.
Once scraping mode is enabled, you will be able to select which value on your screen to scrape, and once that value is selected you can elect to either confirm scrape (save this scraped value to your workflow) or delete scrape (cancel scraping action).
Alternatively, copying values (e.g., ctrl + C) will also automatically create scrapes
Delete Recording
If you would like to delete the workflow you have recorded, you can select this button.
Move Controls
This dragging button will allow you to move the toolbar around your screen.
<aside> 🚨 A caveat to recording – certain sites receive special permissions from Chrome and block recording. You won’t be able to film on new tabs and any Chrome settings pages, and recording may be finnicky on Google Suite apps (like Gmail, Google Sheets, etc.) – because Chrome gives special permissions to these sites, we will not be able to properly access all the page information. For these, we provide special integration steps, which are currently enabled on a per-customer basis depending on the integration.
</aside>
Once you have recorded your workflow, open the dashboard. Click the “Import Workflow” button, and you should see your new workflow listed under processing status.
During this processing period, we are running advanced image models to break down your workflow. We use a combination of vision models and LLMs to analyze the recording of the workflow, break it down into executable steps, and map it into a ML representation. We are also mapping out the site, determining action effects, possible variable inputs, circumventing bot detection, and more.
You will be notified on Slack / Email when your recorded workflows have finished processing. Depending on the length and complexities of the workflows, this could take anywhere from a few minutes to a few hours.
<aside> 👉 Keep in mind – workflows take some time to process after recording, and the longer the workflow, the longer you may need to wait. That said, we notify you ASAP when things are done.
</aside>
Once your workflow is created, you need to review the actions that were detected and configure them as needed.
While reviewing, confirm that all the steps and actions look correct! Please flag anything that looks off – we will promptly look into it.
Note that you can assign any action to be handled manually. If selected, the automation will pause the run at that point in time and request your input manually.
You can also skip any action, which allows you to ignore any extra actions that may have been accidentally included in the recording.
Once your workflow is created, you may also need to map in data that will be used in your automation.
There may be some inputs to the workflow that could change between executions of the workflow.
For example, let’s say you want to check tax information for many countries. Considering that you already recorded the workflow previously, you could now replace the hard-coded country name by a variable, so then you can execute the workflow multiple times to cover your whole list without having to rebuilt.
You can assign or create a variable in an action that can take on values - for example, an input or a multiple choice action. Edit the action and link in a variable via the + button on the right, or via more advanced settings in the three dot menu at the top right.
https://app.arcade.software/share/ZoILal9dkoeX5O7X6PBG
While a variable is being edited, you can apply an LLM transformation on the value to shape it in your desired manner.
On the page to edit a variable, select the wand icon, type your prompt, press the airplane to run the prompt, and apply the transformation if the test output looks right.
For example, you might be filing a form that has multiple input fields for address (ex. address, city, state, zip code), but the address you have on file is contained in a single field (ex. 1 Broadway, New York, NY 10004).
To transform your data to fit the inputs you need, such as zip code, you can run an LLM transformation with the prompt: “Extract the zip code from this address”.
Alternatively, you can transform scraped variables as well. You may scrape a large block of text, but only need a portion of that text or a summary of the text as an output. In the event that you would like to summarize the scraped output, simply create a transformation with the prompt: “Summarize this text”.
https://app.arcade.software/share/U06333wXhkWObxvnT2fV
Creating variables from other variables (e.g., scrapes)
In certain cases, it may be beneficial to create multiple “versions” of a variable.
For example, if an address is scraped and then stored in an “Address” variable (example: “404 Test Street, Circuit City, NY 10101,”), you could transform that entire Address variable, but you may alternatively wish to separate it into distinct variables for Street Address, City, State, and Zip Code.
This can be configured using the Workflow Variables section within the variable editor.
https://app.arcade.software/share/Lng2wJnw0sx98f2qVVC6
Many workflows require business logic overlaid on top of the original workflow.
We allow for:
<aside> 💡 You need to follow each branch of your workflow until the end, to let our system know what has to be done. A good practice for doing that is to record the “main branch” of your workflow, and then build other branches on your own via adding recordings (more on that below) or approving suggested error branches when the workflow runs into new situations.
</aside>
Standard Conditional Logic:
https://app.arcade.software/qpNIeo568bp9ovhElsAj
<aside> 🚨 The conditions “is” and “is not” are case sensitive, so you have to send it exactly as you have configured.
</aside>
AI-Based Conditional Logic:
https://app.arcade.software/share/I92w2egCXSjLQCyfzB9o
For complex actions or actions that aren’t yet supported, you can provide us instructions or requests so that we can still handle them in your workflows. We’ll use our advanced LLMs to perform the action, or we can set up a manual solution for you depending on volume.
The “add recording” step lets you iteratively add to a workflow by recording more portions. This helps to correct mistakes during the original workflow (addressing any missing pieces), record new branches (such as other logical paths on the websites), and break down recording of a long workflow into smaller pieces.
https://app.arcade.software/share/XSe5nuNnCvjFsp3MwV2k
The stop step allows you to set the ending of your workflow.
By default, if you don’t set the stop, we’ll mark the execution as successful upon completion.
Use the interactive demo below to see how it works. In this example, we’ll create a conditional step and will mark the workflow as successful if the country that was input was Italy and as a failure if the country input was anything else.
https://app.arcade.software/share/EZ2q6k6aGGpi0q4d0Xbs
In the above example, when you access the details of the workflow, you’d be able to see requests marked as successful and others marked as failed, depending on the country received.

You may also configure retry settings.
These settings are used if you have a retry step in your workflow.