Self-Attest to a Bill of Materials

Introduction

Creating a Bill of Materials (BOM) is a critical step in documenting the components that make up an AI System. It provides visibility into key elements such as technologies, models, datasets, and infrastructure, ensuring transparency and facilitating compliance. This guide will walk you through the process of self-attesting to a Bill of Materials, allowing you to manually input and verify essential components used in the development, deployment, and management of your AI System. From technologies and models to infrastructure and datasets, this step-by-step guide will help you confidently build and manage your AI BOM.

Navigate to the Bill of Materials Wizard

To create a Bill of Materials, start by navigating to the Bill of Materials Wizard. In the navigation drawer, click the Artifacts section, then select Bill of Materials. Click the blue Add Bill of Materials button to launch the Bill of Materials Wizard.

Input BOM Details

In this step, provide the key information about your Bill of Materials. Start by entering a unique and descriptive name that accurately reflects the project or system you're documenting. Follow this by adding a brief but informative description of the AI project—this could include details about the system’s purpose, scope, or key functionalities. Lastly, assign an owner who will be responsible for managing and maintaining the BOM. These details are essential for tracking and managing your BOM effectively, especially as you add more components.

Continue Without Scanning

For this guide, we’ll skip the automated scan option and proceed with manually attesting the components of your AI system.

Technologies Attestation

Now, you’ll need to document the core technologies used in building and deploying your AI models. These technologies could include programming languages, frameworks like TensorFlow or PyTorch, and other libraries or tools that support your AI workflow. Each technology you add plays a significant role in transforming data into intelligent models, so it’s important to be specific. Include version numbers and any compatibility requirements, as well as whether the technology is critical to a particular phase, such as model training or deployment.

Click the large grey and white 'Add Technologies' box to activate the 'Edit Technologies' dialog.
Input an accurate and descriptive name.
Input a short description of the technology component.
Click the blue 'Submit' button to save the technology component.
Click 'Next' to move on to your Infrastructure components.

Infrastructure Attestation

Next, detail the infrastructure that supports your AI System. This includes the hardware resources (such as GPUs or CPUs) and cloud or on-premise infrastructure necessary for training and deploying models. Ensure that you provide specific information on the hardware configuration and any infrastructure optimizations used to enhance performance or resource efficiency.

Click the large grey and white 'Add Infrastructure' box to activate the 'Edit Infrastructure' dialog.
Input an accurate and descriptive name.
Input a short description of the infrastructure component.
Click the blue 'Submit' button to save the infrastructure component.
Click 'Next' to move on to your Models.

Models Attestation

Here, document any pre-trained models or SaaS models that are integrated into your system. This could include custom models you’ve developed, as well as third-party models that enhance your system's capabilities. Ensure that for each model, you note version numbers, compatibility requirements, and any modifications made to the base model. This step is key to ensuring that the transformations applied to your data are transparent and well-documented, providing a clear understanding of how data is processed and turned into actionable intelligence.

Click the large grey and white 'Add Models' box to activate the 'Edit Models' dialog.
Input an accurate and descriptive name.
Input a short description of the model.
Click the blue 'Submit' button to save the model.
Click 'Next' to move on to your Datasets.

Datasets Attestation

Datasets form the foundation of any AI System, making it critical to provide detailed information on the datasets used for training. Document each dataset, including its source (whether it’s from a database, web scraping, or an internal data lake), its preprocessing techniques, and any augmentation tools used. Also, specify whether the data has been modified or cleaned using tools like Pandas. Providing this level of detail ensures reproducibility and transparency, both of which are essential for managing the quality of your AI Systems.

Click the large grey and white 'Add Datasets' box to activate the 'Edit Datasets' dialog.
Input an accurate and descriptive name.
Input a short description of the dataset.
Click the blue 'Submit' button to save the dataset.
Click 'Next' to move on to Review.

Review

Once all the key components—technologies, infrastructure, models, and datasets—have been attested to, it’s time to review your completed BOM. This step allows you to double-check that each element of your AI System has been properly documented. When satisfied, you can either export the BOM as a JSON file for sharing and compliance purposes or save it within the platform for future updates or reference. This ensures that your AI System’s architecture is fully traceable and ready for audits or further refinement.

Review your completed Bill of Materials.
Click the white 'Export' button to export the BOM as a JSON file.
Click the blue 'Add BOM' button to save the Bill of Materials.

Did this answer your question?