Build and deploy an AI App

You can create a new AI app by either starting with a pre-built template or building one entirely from scratch.

1. Build an AI App from a Template

On the AI Apps page, click on Build an App.

You can now see all the templates available on the MarkovML platform. To ease your process, you can use the search bar to look for a desired template, or the navigation on the left of the page to view different use-case-based groups.

Click on the template that you want to use.

When you click on a template, you will get a pop-up that shows the description and flow of the template, along with similar templates for the same use case. If this looks like a good match, clickUse this Template.

This opens the App Builder screen, with pre-filled configuration for inputs, operations, and output. Based on your requirements, you can make the necessary edits to the input and operations. There is also a wide range of input types and operations to choose from if you change the flow of the app.

For example, you can change the pre-configured input to add to your own dataset. This can be a .csv file from your computer or data from sources like AWS S3, PostgreSQL, or Snowflake. You can also change the remaining steps, like operations to edit/filter your data before processing, and even change the LLM prompts to fit your requirements better.

Ensure that all the blocks are configured in the right sequence to get the desired output.

Once you have the desired flow of steps, click on Deploy App. This shows you a preview of the app page.

Click on Edit the app view to further customize your app. You can change the field names, descriptions, and also the title of the Run the app button.

Finally, click Deploy App to confirm your changes and deploy the app. Your new app will now be available in the AI app library.

2. Create an AI App from Scratch

On the AI Apps page, click on Build an App.

Click on Use blank canvas to start building an app from scratch.

This opens the App Builder screen. You can now create a sequence of actions using the input types, operations, and output types provided.

For example, you can choose the type of input as a .csv file from your computer or data from sources like AWS S3, PostgreSQL, or Snowflake. Then, you can add filter operations to filter your data before processing, and even add/edit the LLM prompts to fit your requirements. Also, choose where you would like the results to be saved—whether it’s an S3 bucket, your computer, or directly within the MarkovML library.

Ensure that all the blocks are configured in the right sequence to get the desired output.

Once you have the desired flow of steps, click on Deploy App. This shows you a preview of the app page.

Click on Edit the app view to further customize your app. You can change the field names, descriptions, and also the title of the Run the app button.

Finally, click Deploy App to confirm your changes and deploy the app. Your new app will now be available in the AI app library.