Citadel Enterprise Solutions, end-to-end AI quality consulting, is now available

Citadel AI Inc., the leader in improving AI performance, is launching Citadel Enterprise Solutions, our end-to-end consulting service for improving AI quality.  

Along with Citadel Enterprise Solutions, we will also begin trial deployments of Citadel Lens (alpha), an automated AI verification tool, in addition to Citadel Radar (beta), an automated AI monitoring tool, which was released this May.

The challenge of maintaining AI quality

Unlike the ideal world in a research paper, the real world is constantly changing, which degrades the performance of your AI system over time. For example, external events such as the COVID-19 pandemic caused demand forecasting models to produce unreliable predictions. Proactively detecting errors and maintaining prediction quality, before they result in business losses and compliance issues, is critical to the success of enterprise AI. However, it’s a significant challenge (in terms of cost and resources) for most companies to have data scientists constantly monitoring AI systems during deployment.

Our end-to-end solutions from AI verification to monitoring

Citadel Radar (beta) is currently in trial use by major SI companies, manufacturers, and AI startup companies. By combining Citadel Radar, our quality monitoring tool, and Citadel Lens, our new quality testing tool, you can automatically detect AI issues from model training to model serving, with a visual user interface. Our products accelerate the automation of maintenance and operational tasks, and protect enterprise customers from new risks embedded in AI systems.

In addition, we are launching “Citadel Enterprise Solutions”, our consulting service to solve issues related to your AI performance. By leveraging our suite of products, Citadel Enterprise Solutions can easily detect problems from testing to deployment, providing end-to-end solutions to maintain and improve the quality of enterprise AI for individual customers.

‍

[Citadel Enterprise Solutions] End-to-End AI Quality Consulting

• Identify quality issues through our continuous monitoring tool, Citadel Radar
• Our team proposes improvement measures in terms of data quality, model quality, etc.
• Verify and test updated model versions through Citadel Lens

[Citadel Radar] Automated AI Quality Monitoring, during model serving

• Proactively detect, block, and visualize errors before they impact your business
• Sophisticated single-package solution to streamline maintenance and operational tasks
• No need to modify your AI, works out-of-the-box with any model or framework

[Citadel Lens] Automated AI Quality Testing, during model training

• Automatically detect and compare low-performing customer segments
• Detect new errors and incompatible predictions between different model versions
• Comprehensive reliability testing, such as confidence calibration, feature sensitivity, and error analysis

‍‍Increasing demand for AI quality assurance and accountability

As society grows increasingly reliant on AI systems, governments and enterprises across Europe and the U.S. are focusing their attention on the accountability and fairness of AI. For example, the EU announced proposed regulations on the use of artificial intelligence in April 2021. 

In order to build responsible AI systems, there is a growing movement to publish ethical principles and operational playbooks for AI, mainly from technology companies such as Google and Microsoft. Detecting AI errors in real-time, updating models & data appropriately, and maintaining trust is critical to enterprise AI systems.

Citadel AI Inc. is a startup based in Tokyo and received a seed grant from The University of Tokyo’s IPC 1st Round in 2021. Our vision is to provide you with AI that you can always trust. We build solutions that efficiently solve AI reliability issues through automation. Our co-founder was the former product manager for ML infrastructure at Google Brain, and is leading Citadel AI’s product development.

About Citadel AI Inc.