Overview of IBM watsonx.ai and IBM watsonx.governance software

Tools for preparing data and working with models

Your watsonx.ai tools are in collaborative workspaces called projects.

You can use watsonx.ai tools to prepare data and work with models in the following ways:

• Prepare data: Refine and visualize your data files or data tables in remote data sources or generate synthetic tabular data.
• Build machine learning models: Automatically generate model candidates, create machine learning model training flows, or write model training code in Python or R.
• Work with foundation models: Experiment with generative AI prompts, tune foundation models for your use case, or write generative AI solution code in Python.
• Automate model lifecycles: Create repeatable and scheduled flows that automate notebook, Data Refinery, and machine learning pipelines.

For a list of tools, their levels of automation, and whether you can use them to prepare data or work with models, see Analyzing data and working with models. For generative AI tools, see Developing generative AI solutions.

Deployed foundation models

IBM watsonx.ai has a range of large language models for generative AI that your administrator can deploy in your cluster. The deployed foundation models include open source models from Hugging Face and IBM foundation models. You can also deploy your own custom foundation models. You can customize foundation model behavior by tuning a foundation model.

For a list of included foundation models that your administrator can deploy for watsonx.ai, see Supported foundation models.

Usage resources for computing and inference

When you run tools or your AI solutions with watsonx.ai, you consume the following types of resources:

Compute usage

When you run jobs, notebooks, experiments that train or tune models, or deployments, you consume compute resources. Compute resources include the appropriate hardware and software that are specific for the workload. Your compute resources can vary depending on which services are installed on your cluster. See Runtime environments.

Inferencing usage

When you run inferencing against foundation models, your inferencing usage depends on the number of the tokens in the prompt input and output text. Tokens are basic units of text. Inferencing is run on GPU hardware.

Tools for governing models

You can use watsonx.governance tools to govern models in the following ways:

• Monitor and evaluate models: You can monitor model output and explain model predictions. Your watsonx.governance model monitoring and evaluation tools are in projects and deployment spaces.

• Track and document AI use cases: You can view model lifecycle status, general model and deployment details, training information and metrics, and deployment metrics. Your watsonx.governance model tracking tools in AI use cases.

• Manage governance activity from the Governance Console Optionally integrate with the watsonx Governance Console from IBM OpenPages. Sync data from factsheets with the Governance Console, and extend governance capabilities with workflows and other compliance tools.

Connectivity

You can create connections to remote data sources and import connected data. You can configure connections with personal or shared credentials. For a list of supported connectors, see Connectors.

You can share connections with others across the platform in the Platform assets catalog.

Administration

IBM watsonx contains the core of IBM Cloud Pak for Data and has the same administration features and experience. Administrators can perform the following types of tasks:

• Installing, upgrading, or migrating the software
• Backing up or restoring the software
• Monitoring the platform
• Securing the environment
• Auditing events
• Forwarding alerts, notifications, and announcements
• Setting up services
• Managing resources
• Managing users

See Administering Cloud Pak for Data.

Storage

IBM watsonx and Cloud Pak for Data require a persistent storage solution that is accessible to your Red Hat OpenShift cluster. All the assets that you create with watsonx.ai and watsonx.governance are stored in that persistent storage solution.

See Storage requirements.

Workspaces

Watsonx is organized as a set of collaborative workspaces where you can work with your team or organization. Each workspace has a set of members with roles that provide permissions to perform actions.

Most users work with assets, which are items that are created or added to workspaces by users. Assets can represent data, models, or other types of code or information. Data assets contain metadata that represents data. Assets that you create in tools, such as models, run code to work with data. You can also create assets that contain information about other assets, such as model use cases that contain metadata, history, and reports about models. See Asset types and properties.

You can work in these types of workspaces in watsonx:

• Projects
• Deployment spaces
• Platform connections
• AI use cases

You can search for assets across all workspaces that you belong to.