I like that IBM watsonx.governance provides us with a unified, policy-controlled environment for our entire client AI portfolio, offering a clear picture for model risk management. This feature is essential as it helps us manage all AI assets consistently, whether the model is a machine learning model or a new-generation AI large language model. Also, the best part is being able to integrate any large language model, from Salesforce, AWS, etc. By enforcing client standards through AI policy packs, we're able to keep everything under compliance and tremendously reduce risk. Review collected by and hosted on G2.com.

I don't like the initial configuration complexity and the friction it sometimes causes when integrating with non-IBM ecosystems. This is better with Azure and AWS, where proper documentation is found and configuration is pretty simple and straightforward. Also, I believe they could enhance user adoption by simplifying the out-of-the-box connectors and providing low code templates for multi-cloud deployments. Review collected by and hosted on G2.com.

I appreciate how IBM watsonx.governance serves as an "AI copilot" for compliance. It streamlines the often tedious paperwork, such as factsheets, and actively identifies potential bias and risk. By ensuring that all AI models adhere to global regulations automatically, it enables businesses to scale trustworthy AI in a safe and efficient manner. Review collected by and hosted on G2.com.

I dislike that the platform has a steep learning curve due to its complexity and that its high cost can be a barrier for smaller businesses.

Users also report occasional struggles with integrating it smoothly with non-IBM tools and a need for better documentation and tutorials. Review collected by and hosted on G2.com.

I really like the automated AI Factsheets feature of IBM watsonx.governance because it creates clear, audit-ready documents without any manual work, making compliance much easier. The real-time monitoring of bias and model drift is also a standout for me as it helps catch issues early, reducing risk and ensuring the models stay reliable and stable in production. Also, it integrates well with the model development platform, data pipelines, and monitoring tools we already use. Overall, these features reduce ongoing manual efforts and provide a smoother experience. Review collected by and hosted on G2.com.

I think the setup and initial configuration can feel a bit complex and time-consuming, especially when integrating with existing tools and workflows. Review collected by and hosted on G2.com.

I really appreciate IBM watsonx.governance for several compelling reasons. One of the standout features for me is the AI Governance & Trust capabilities, which enhance the reliability and ethical deployment of AI models, ensuring that they can be used with confidence across my organization. The flexibility of the foundation model is another highlight, providing me with the adaptability needed to tailor AI applications to diverse scenarios and requirements effectively. I also find the enterprise integration capabilities to be highly beneficial, as they allow seamless connectivity with other tools and systems within our existing infrastructure, which simplifies the workflow and boosts efficiency. Additionally, I value the pre-built models and templates that IBM watsonx.governance offers, as they significantly streamline the initial setup process. It's also incredibly convenient to have integration with the broader IBM ecosystem, such as IBM Cloud Pak for Data, which enhances my overall operational capabilities. These features collectively make IBM watsonx.governance an invaluable tool in building, training, and deploying AI models, especially when managing data at scale. Review collected by and hosted on G2.com.

I think IBM watsonx.governance could benefit from a broader model ecosystem. Expanding this would make the platform more versatile. Adding more support for open-source, community-driven models and domain-specific models would greatly enhance its flexibility and adaptability. Review collected by and hosted on G2.com.

The tool's proposal is good. The user experience as well. The content it aims to present. Review collected by and hosted on G2.com.

Difficult to use.

Limited in connections.

Limited in forms of implementations.

The SDK is complex to use and has the wrong version in the Pypi repository.

High latency and connection breaks with OpenScale during tool usage.

Not very debuggable, difficult to find out what is going wrong when an error occurs. Review collected by and hosted on G2.com.

It helps maintain control over the AI we use, especially in an environment where sensitive data is handled. Review collected by and hosted on G2.com.

Maybe as a team we are not getting the most out of it. Review collected by and hosted on G2.com.

I think it's a very necessary service to have to babysit and help regulate AI usage. Review collected by and hosted on G2.com.

The price point is a bit high, and it would be easier to implement it fully if it were slightly cheaper. Review collected by and hosted on G2.com.

I use IBM watsonx.governance to make sure all my AI models are compliant with regulations and transparent. It has been a game-changer in solving problems related to AI compliance and risk. I really like the automated AI Factsheets, which provide perfect, audit-ready transcripts for every model I deploy. The ability to automatically monitor bias and drift in real-time is a huge time-saver for my team, eliminating the need for countless hours of manual audits. Review collected by and hosted on G2.com.

The initial setup complexity and steep learning curve. It was not easy to setup in initial stages and it was kind of a multi-week implementation project. Review collected by and hosted on G2.com.

I like how watsonx governance combines model risk, generative ai, and governance. For example, the ability to create a use case, track a model, associate with workspaces, and evaluate prompts all in a single area are great. Even better, but IBM recently launched a new feature called the Governance Console where you can track multiple models, enable and create controls, and evaluate risks.

Governance had Gen AI metrics like ROUGE and BLEU, as well as the ability for classification like F1, Accuracy, precision, recall, etc Review collected by and hosted on G2.com.

I don't like how an evaluation in Watsonx Governance overwrites the previous evaluation results if done on the same day. OpenScale will keep the history, but governance in the AI Factsheet does not. Same issue exists for the trend charts.

For example, the trend charts available in OpenScale have a time setting for hourly, daily, weekly, etc. But let's say i do 20 evaluations in a day. Then a week later I want so show how those 20 evaluations did, and the overall trend. If I change to hourly, then the graph is too sparse (it's been a week), but if I change to daily, then it's just a single dot (all 20 evals were on a same day). Wish OpenScale and Governance would fix this.

Then the ability to show the generated_text against the reference (ground truth) text. Yes, you can download the results to a CSV but there's two big missing features (1) the downloaded CSV is missing the reference text, it only has the generated text. So you have to come up with a way to match that record to its original source if you want to compare, and (2) you have to do this each time for each subset. Why not just have the ability to view that record in the software and avoid having to download the CSV? the text is truncated and there's no way to see it Review collected by and hosted on G2.com.

It monitors AI models for fairness, bias, drift, and key metrics, proactively detecting and mitigating risks Review collected by and hosted on G2.com.

The setup is complex and the learning curve is steep Review collected by and hosted on G2.com.