Belgium’s Soda acquires NannyML to help companies spot AI and data problems earlier

Brussels-based Soda, a data quality and observability platform, has acquired Leuven’s NannyML, a company focused on AI monitoring and observability. 

The acquisition integrates NannyML’s AI technology into Soda’s platform. The combined system aims to detect issues earlier, respond to changes, and reduce irrelevant alerts.

Founded by Hakim Elakhrass, Wiljan Cools and Wojtek Kuberski, NannyML is a monitoring tool for AI systems in production. It helps teams detect performance changes, data drift, and model failures without using ground truth labels. 

Used by data science teams across various regions, NannyML supports post-deployment model monitoring and performance estimation. It is an open-source Python library that works with tabular data, classification, and regression models. 

The company supports model performance estimation, business value estimation, data quality checks, and multivariate drift detection. The tool is model-agnostic and includes an interface with interactive visualisations.

Aim of the acquisition

Enterprises are managing growing data complexity, including larger data volumes, fragmented pipelines, and AI systems in production. This creates a need for data quality tools that can adapt to evolving conditions.

Soda’s acquisition of NannyML brings AI-driven capabilities into its platform, enhancing data quality workflows and reducing blind spots and false positives.

Maarten Masschelein, CEO and co-founder of Soda, says, “With NannyML, we are accelerating our ability to deliver intelligent data quality infrastructure that adapts to the way modern data systems work.”

“This is a strategic step in our long-term vision: to give data teams full visibility from ingestion through to automated decision-making, so they can move fast without compromising trust, making Soda a foundational component of AI infrastructure.” 

Soda is continuing its work on an AI-first data quality platform while using the NannyML acquisition to support a broader goal of full-stack observability. This approach spans raw data, transformations, model outputs, and autonomous agents. 

Hakim Elakhrass, co-founder and CEO of NannyML, adds, “At NannyML, we set out to help data teams monitor what matters most, model performance in production, without drowning in false alarms.”

“Joining Soda allows us to carry that vision further. Together, we are creating a unified platform that ensures trusted, reliable outcomes across both traditional analytics and AI Systems.”

Brief about Soda

Founded in 2018 by Maarten Masschelein and Tom Baeyens, Soda is a platform for data quality and observability. It supports data teams in preventing, detecting, and resolving issues across data pipelines. 

Soda is used by organisations to manage data products at scale and combines testing, monitoring, and collaboration in a single platform.

The company was created in response to challenges its founders faced while working on data teams, where issues often remained hidden until they caused downstream problems. Soda was built to help teams identify and address data issues early, reduce disruptions, and enable consistent data use across organisations.