Create an Agentic Tool

The Agentic Tool feature in the Data Science Lab allows users to configure and launch custom environments for developing, training, and executing AI/ML workflows. Administrators and data scientists can allocate compute resources, choose frameworks, integrate Git repositories, and manage project settings efficiently.

Steps to Create an Agentic Tool

• Navigate to Agentic Tools > Create > Create Agentic Tool.

• The Create Agentic Tool form opens.

Fill in Project Details

• Project Name*: Enter a unique name for the Agentic Tool project.

• Project Description: Optionally, describe to define the project purpose.

Configure Environment and Algorithms

• Select Algorithms: Choose the algorithms to be applied from the drop-down list. Currently supported algorithms are Regression, Classification, Forecasting, Unsupervised, and Natural Language Processing.

• Environment*: Select the desired environment. Currently supported environment is Python.

Allocate Resources

• Resource Allocation*: Select the resource level for CPU/GPU and memory. Options include:

  • Low

  • Medium

  • High

• Idle Shutdown*: Specify the idle time limit after which the notebook session disconnects automatically if inactive.

Add Libraries and Repositories

• External Libraries: Upload or define any external libraries needed for your project.

• Git Repository: Provide the Git repository URL if code integration is required.

• Git Branch: Select the Git branch to be synced with the project.

• (Optional) Sync Git repo at project creation: Check this box to automatically pull code from the repository during setup.

Configure Compute Settings

• GPU Type: Select a GPU type if required for your workload (default: None).

• Nodepool: Choose the nodepool for execution (e.g., execpool).

Finalize

• Click Save to create the Agentic Tool.

• Click Reset to clear all fields and start again.

Key Notes

• Resource Allocation: Defines the compute capacity for the notebook container.

• Environment: Preloaded frameworks allow seamless execution without manual package installation.

• Idle Shutdown: Automatically deactivates the project after inactivity; reactivate to continue work.

• Git Integration: Ensures code versioning and reproducibility.