McKinsey acquires AI development startup Iguazio

AI

Management consulting firm McKinsey & Co. today announced that it has acquired Iguazio Ltd., a startup with a platform for automating artificial intelligence development projects.

The terms of the deal were not disclosed. Iguazio previously raised $72 million from Samsung Electronics Co. Ltd. and other investors.

McKinsey is one of the largest players in the management consulting market. The company, which doesn’t disclose financial data because it’s privately held, is believed to generate annual revenues of more than $10 billion. It has more than 30,000 employees in 60-plus countries.

Tel Aviv-based Iguazio, in turn, sells a software platform of the same name that helps companies develop AI models in a more automated manner. The startup counts NetApp Inc., Robert Bosch GmbH and other large enterprises among its customers.

“After analyzing more than 1,000 AI companies worldwide, Iguazio was identified as the best fit to help us significantly accelerate our AI offering – from the initial concept to production, in a simplified, scalable and automated manner,” said McKinsey Senior Partner Ben Ellencweig.

Iguazio’s platform enables companies to create training datasets, use them to develop a neural network and then quickly deploy that neural network to production. Historically, the AI development process involved a significant amount of manual work. The platform promises to automate much of that manual work for software teams. 

AI applications often don’t analyze the data they ingest in its original form, but simplify it first. A neural network that takes product prices as input, for example, might round each product price to the nearest whole number before carrying out analysis. The simplified data points that are produced through this process are known as features.

When an AI model ingests new records, it must turn them into features in near real-time to avoid holding up processing. In practice, however, achieving such real-time performance is often technically challenging. Iguazio’s platform includes features that promise to simplify the task for enterprises and thereby improve the performance of AI applications. 

Another challenge the startup promises to ease is the task of addressing data drift. That’s the term for a technical issue AI models encounter when the data they process changes over time. A neural network trained to detect technical errors in a certain kind of industrial equipment, for example, might encounter data drift if it’s repurposed to analyze error logs from a different type of machine.

When such data drift errors, the accuracy of AI models often declines. Iguazio’s platform can detect data drift and automatically retrain AI models to address potential accuracy issues. The startup’s platform also generates dashboards that a software team can use to track how the reliability of its neural network changes over time. 

Alongside its namesake platform, Iguazio offers two open-source AI development tools. The first tool, MLRun, automates the process of packaging AI software code into software containers. Iguazio is also the developer of Nuclio, which makes it easier to process the data that companies use in AI projects.

Following the acquisition, McKinsey plans to integrate Iguazio into its QuantumBlack professional services unit. The unit helps companies develop and implement AI models. QuantumBlack was formed in 2015 after McKinsey bought a London-based analytics and machine learning company of the same name.

Photo: Iguazio