Veritable Monitoring
Veritable Monitoring goes beyond basic machine learning performance, helping data science and ML Ops teams easily oversee and fix all of their models.
Tired of the wild goose chase?
Experience lightning-fast, complete visibility of your models
Based on 6+ years of research, our approach is highly accurate, eclipsing current tools
Easily comply with artificial intelligence guidelines and regulations
“We are excited about a solution that goes beyond current ad hoc tools and allows us to not only quickly and effectively identify emerging issues, but also debug and resolve them.”
HarperCollins Publishers
Monitor and troubleshoot model performance at scale
Get the broadest view
See all of your models in one place. With a set of analyses that nobody else offers, Veritable Monitoring provides a more complete view, including:
- Consequential detection of input and output drift
- Quick diagnostics of priority segments
- Model accuracy tracking, even without data labels
- Bias tracking
- Data quality problems
Understand why
Go from metrics to root causes, fast. Unlike other solutions, which just tell you what’s going on, Veritable also pinpoints the underlying drivers of those metrics.
- What’s the specific cause of model instability?
- Which segments are most impacted?
- Which features are degrading model performance, and how much?
Ensure fairness and compliance
As algorithms increasingly influence people’s lives, minimizing model bias is an imperative.
- Monitor model bias for protected classes or any identified class
- Identify which specific features are contributing most to bias results, for faster explanation and resolution. This deep analysis is unique to Veritable.
Get notified
Veritable ensures that you can rapidly respond to any emerging incidents:
- Configure alerts for any Veritable monitoring diagnostic metric, such as model stability, feature influence, and feature drift
- Send notifications to any user across nineteen different channels, including email, Pager Duty, Webhooks, and Slack
- Use annotations to track reasons for alerts and anomalies or to document resolutions
Easily deploy and scale
Veritable embeds easily within your existing infrastructure and workflow
- Deploy on premises or in your cloud, including private cloud, AWS, Google, or Azure
- Integrate easily with popular model development and model serving solutions
- Scale to meet high model volumes
- Export data via APIs to BI tools such as Tableau and Looker
Resources
