Last last year, we launched watsonx.governance on IBM Cloud with our introduction of the Essentials Cloud Plan. Today I am excited to announce our next major release that supports AI Anywhere.

AI Anywhere means two things:

1. Monitors any AI models
2. Installs Anywhere

In addition, we are pleased to announce powerful additional features for Governance Risk and Compliance featuring workflows with approvals and dashboards, supporting the management of AI.

Learn more about the latest features in watsonx.governance at our webinar, March 12th at 11 AM ET, register here.

AI Anywhere = Monitors any AI models

First, AI Anywhere means we can monitor any AI models from any vendor. This means whether you are building and deploying models on watsonx.ai, Amazon Sagemaker/Bedrock, Google Vertex, Microsoft Azure or any other vendor, watsonx.governance can now govern these deployments. This works for both Predictive ML and LLMs.

For LLMs, watsonx.governance can calculate and alert if thresholds are breached for quality metrics for Text Summarization, Text Classification, Entity Extraction, Content generation, and Q&A use cases. We can also calculate and alert for instances of Personal Identifiable Information and Toxic language found within prompts. Also for RAG Q&A use cases, we can attribute output back to the reference or context data provided for validation and understanding. Each of these calculations can be done for any vendor at development time. (Details in our documentation here.)

For Predictive ML, both development and runtime metrics can be monitored for any model including quality metrics, drift, and fairness monitoring. In addition, explanation of model transactions provides true understanding of how ML models operate either in development or production no matter where you are building or deploying your model today.

AI Anywhere = Installs Anywhere

AI Anywhere also means watsonx.governance can be installed anywhere. watsonx.governance is now available for installation on premise, or on any customer managed public or private cloud environment. Because it is built on Red Hat OpenShift, it is flexible on a variety of environments. This way, enterprises can install on any public or private cloud of their choice.

This supports the true openness of our watsonx platform where customers can choose the deployment model that is right for them, including the ease and simplicity of IBM Cloud or the flexibility of the cloud of their choice.

New Governance Risk and Compliance functionality

Lastly, a major announcement: watsonx.governance now includes the ability to manage risk with integrated governance risk and compliance functionality. We now offer:

• Out-of-the-box dashboards that provide a broad view across use cases and risks
• Out-of-the box workflows that can easily be customized for implementation of your processes, standards, and best practices
• UI-driven/low code functionality that make it easier to onboard new (and non-technical) users

This provides:

• Easier collaboration across different stakeholders — policy / risk teams, ML/LLMOps teams, data scientists and model validators
• The power to visualize risk across traditional ML and GenAI
• Reduction in overhead when using manual processes and enables scalable implementation of best practices
• Accountability and oversight of enforcing policies and regulations
• Use cases to map model risk to broader operational risk

This essential part of AI Governance is now available on our OpenShift version today.

Learn More and Try Today

To learn more about watsonx.governance or give it a try for free, visit our website at https://www.ibm.com/products/watsonx-governance

Here you can start a trial or setup a briefing where you can request a demo of the above features in action.

#watsonx.governance