Integrating AI Capabilities into Automation

This document outlines the process of enhancing an existing automation program by integrating AI capabilities. The instructions focus on adding a Merlin connector to handle edge cases and automate error processing within the automation. By following these steps, users can improve the efficiency and accuracy of their automation tasks.

Step 1

Begin by ensuring the existing automation program is loaded. Next, incorporate AI capabilities by adding a Merlin connector into the automation. This addition will allow for handling edge cases efficiently.

Step 2

The Merlin connector will process requests based on the provided context, effectively managing errors and determining subsequent actions. In the current automation setup, there is a default branch that generates a failure notification to the team.

Step 3

Modify the default scenario by routing unmatched locations through the Merlin connector. This connector will process the request and facilitate the regular workflow if applicable.

Step 4

If the request cannot be processed, an email notification will be sent. Begin this by adding a new step to include the Merlin connector.

Step 5

Search for 'Merlin' and review the available options. Select 'Merlin Text Generation' and click to proceed.

Step 6

After adding Merlin, you'll have options to generate or summarize text. Select 'Generate Text' and choose the Claude-Helio model.

Step 7

Provide detailed instructions to the Merlin connector to ensure it correctly processes requests and generates the desired output format.

Step 8

Adjust the token count to 2,000 and set the temperature to 0.5. Define the purpose of the operation clearly.

Step 9

Review the instructions, specifying that given row, location, and environment details, the Merlin connector should consult a device policy catalog and return the relevant information. Include the catalog JSON as input.

Step 10

Copy and input the JSON-formatted data, which includes details on row, location, environment type, and device allocation requirements. Ensure adherence to specified rules for output generation.

Step 11

The output should be in JSON format or an error value if no match is found. Run the workflow to test its functionality.

Step 12

Observe the output, which will display the required device, VPN, and application information in JSON format.

Step 13

Introduce a Boolean condition to verify that Merlin produces a result. If the response is not null, proceed to parsing.

Step 14

Add an object helper and utilize the string design.

Step 15

Parse the response and ensure it corresponds with the test.

Step 16

Use the resulting ID appropriately within the process.

Step 17

Transfer the relevant content from the previous step to the current process.

Step 18

Incorporate additional information as necessary, utilizing the panel for data integration. Reference previous steps for context.

Step 19

Insert the required state information.

Step 20

Finish by copying the data to the appropriate location and including an alternative team's email address as needed.

Step 21

Update with your ID and prepare the setup for actual operation.

Step 22

Modify the snapshot to include team data and ensure real-time operation is enabled.

Step 23

Add necessary contact information, ensuring the system bypasses the default path when configurations don't match, preventing errors.

Step 24

The Merlin connector now effectively manages situations based on provided context, facilitating standard onboarding processes or generating responses accordingly.

Step 25

In cases where processing is ineffective, the system defaults to an error-handling flow. By leveraging AI, specifically through Merlin, edge cases are managed effectively, enhancing the automation process.

Step 26

An alternative approach involves providing context within agent workflows for the Merlin agent to manage scenarios. This aspect will be covered in future documentation.