OpenAI & AppSheet #2: Create Dynamic Starters for Continued Conversations in Your ChatGPT Interface

Hello OpenAi community!

I’m excited to share the second video in my ongoing series that teaches you how to create a ChatGPT-like interface in AppSheet.

In this video, we’ll learn how to expand the app we built in the first part by adding dynamic starters. If you missed the first video, you can catch up here.

Watch the second video below

In this tutorial, you’ll learn how to add the ability to dynamically change the first bit of text in the prompt sent to OpenAI, which I call “Starters.” These starters can serve multiple purposes, but our main goal is to use them for providing instructions to the AI system.

This setup allows you to easily switch between different instruction sets, making it incredibly simple to A/B test instructions.

Here’s a quick overview of the steps we’ll follow to update the app from part 1:

  1. Add a new table to the system called “Starters” with the following columns:

    1. Starter_Title
    2. Starter_Text
    3. Starter_Tokens
    4. Starter_ID
    5. CreationDateTime
  2. Add a new column to the “Messages” table as an Enum (base type REF to the Starters table), allowing users to select the Starter they want to use.

  3. Modify the Automation Bot by replacing the hard-coded instruction set in the Concatenate() in the “Get Response” task’s [Prompt] parameter with a dereference from the selected Starter: [Messages_Starter].[Starter_Starter_Text]. (No modification to the script is necessary.)

By the end of this video, you’ll have an even more powerful ChatGPT-like interface in your AppSheet app that can handle multiple instruction sets and adapt to your needs

I hope you find this tutorial helpful! Don’t forget to like, comment, and subscribe to my channel for more updates in this series. If you have any questions or suggestions, feel free to leave a comment below.

Happy app building! Small moonwalker.gif