Spark, the AI Script Assistant

Overview

Spark is our AI Script Assistant that provides editing support on scripts by making Mainstay’s decade of research and expertise easily accessible to anyone. Launch Spark from the bottom right-hand corner of the Script Builder and decide how you’d like to partner with Spark while working on your scripts. 

Working with Spark, the AI Script Assistant

Script Builder

When working on a script in the Script Builder, Spark can be launched by clicking on the Mainstay spark icon in the lower right-hand corner of the page: 

Once launched, Spark can Rephrase, Revise, or Rephrase & Revise your script: 

• Rephrase – For Rephrase, Spark will rewrite the script to match the tones and traits of your bot as they are set up in the Bot Personality Controls. If you do not currently have a personality configured, you will want to set one up before using the Rephrase functionality. 
• Revise – With Revise, Spark will review the script using Mainstay’s SPARK Model, behavioral science techniques known to move people to action, and best practices we’ve developed over the past decade for creating impactful and engaging SMS content. Spark will suggest changes to the script that utilize these principles and that can help make the script more impactful and engaging for learners. 

• Rephrase & Revise – Selecting Rephrase & Revise and Spark will do both in one review. 

  

  Once Spark has finished working on the script, you will be redirected to a review page where you can review each of Spark’s suggested changes. 

Spark's Review Page

On Spark’s review page, the original script and a script with Spark’s suggested changes will be presented in a side-by-side view so that you can easily see the proposed changes. If you are confident with Spark’s changes and do not want to go through each message individually, you can click the ‘Accept all’ button in the upper right corner and then click ‘Apply changes’ in the bottom right corner and Spark’s suggested changes will be applied to the script. 

However, if you’d prefer to review each message individually, you can scroll through the scripts side-by-side and review each suggested change.

In the original message on the left side, any content that Spark suggests removing will have strikethrough formatting. In the suggested message on the right, any new content that Spark suggests is highlighted so that it is clear what Spark recommends adding.  

If you’d like to better understand why Spark is making these suggested changes check out Spark’s Coach Note attached to each set of side-by-side messages. A high-level overview is provided; however, if you’d like more information, click the ‘read full note’ button and the Coach Note will expand. 

In the expanded view there are two additional pieces of information: 

• What Changed – This is a list of the ‘why’ behind Spark’s suggested changes. There are four potential categories into which changes can fall: 
  • SPARK – This section shares how one of the five dimensions of SPARK is being applied in the suggested changes. It will call out the specific dimension (e.g. Autonomy-Supporting) and then provide a short explanation of how it shows up in the changes. 
  • Behavioral Science – This section highlights any specific behavioral science techniques being used in the suggested changes. It will call out a specific behavioral science technique (e.g. Chunking) along with a short explanation of how it shows up in the changes. 
  • Best Practices: Design Principle – This section highlights one of the best practices we’ve developed and refined. Spark will explain the best practice as it shows up in the suggested changes (e.g. leading with emotional intelligence before making an ask when a learner admits to not completing a task). 

  • Best Practices: SMS Formatting – This section calls out opportunities that improve readability and deliverability for SMS communication. This could look like adding line breaks to improve readability or avoiding putting links at the end of a message as that is known to trigger spam filters.

     

• Try This Next Time – This section offers an explanation of how you could use the different SPARK dimensions, behavioral science techniques, and best practices Spark has identified in this message in other similar messages in the future. This piece is intended to be used as a teaching tool so that you can utilize these principles in your own writing. It is also a great way to better understand Mainstay’s research and expertise in the context of your specific messaging
   

As you review each message, there are three actions that can be taken:

• Accept – Clicking the ‘Accept’ button will keep Spark’s changes. At the end of the review process, this will rewrite the original message content with Spark’s suggested changes. 
• Reject – Clicking the ‘Reject’ button will ignore Spark’s changes and keep the original message content as is. At the end of the review process, this will essentially delete Spark’s suggested changes and the original message copy will remain. 
• Add as a Comment – Clicking the ‘Add as a Comment’ button will take Spark’s suggested changes and add them as a comment attached to the specific message on the Script Builder page, just like any other Script Comment. This is a great option if you like some of Spark’s changes but not all of them – this way in the Script Builder you can decide what pieces of Spark’s suggested changes you want to copy into the original message and which pieces you’d prefer to ignore.

At any point in the review process you can change your decision for any message by clicking on the ‘change decision’ option.

Once you’ve completed a full review of the script, click the ‘Apply changes’ button in the bottom right corner and these changes will be implemented. 

Returning to the Script Builder

After applying changes, you will be redirected to the Script Builder with changes applied throughout the script. This includes any of Spark’s suggested changes that were added as a comment.

These comments will be found attached to the appropriate message, click the pink message icon and the comment panel will open showing the comment.

From here, you can make any necessary remaining edits to your script and then either tag colleagues for feedback by adding a comment and using @-mentions to bring your team in, or click the ‘Schedule Campaign’ button in the upper left corner to get this script out to your learners.