Automates Feature Engineering and Predictive Analytics

When it comes to developing deep learning predictive models, there are several stages to building a model from raw data. In the first stage, it must be determined if AI can be of use and how, thus, what specific problems can it solve. For instance, a cloud infrastructure company with thousands of servers in deployment, predicting server failure is a valuable metric. For an online company, being able to predict and optimize customer churn may lead to an increase in profit by a few percentage points. For those with sophisticated supply chains and a large inventory of products, models can be used to predict the optimal price point for a product that improves sales and profitability, while keeping its cost structure in check.

In the second stage, the AI engineer must be able to create the perfect data set from dirty data, which is likely to reside in multiple tables and sources, with plenty of missing values, making the ETL process a nightmare. In the third stage, selecting the ideal framework from dozens out there is the next interesting challenge. If it is determined that TensorFlow is the right fit, the process of loading the data set, setting up TensorBoard to visualize outcomes, tuning the right combination of hyperparameters, and debugging any issues that arise is anything but straightforward. Managing the entire process requires great skill, extraordinary focus, and lots of time. Many of the steps of getting from point A to point B are tedious, manual, boring, and time-consuming that can frustrate the brightest of minds, especially for the non-data scientist.

This is where comes in. The AI startup has developed a platform that simplifies the process of building predictive models. They are part of a growing startup ecosystem that enables non-data scientists to build predictive models from raw data in a few simple steps. Some call this market segment AutoML. The value prop for the AutoML startup is simplicity, just load the raw data, configure some parameters, and let the platform do the rest. Thus, all the complexities in building models are done behind the scenes.

In the case of, simply connect the raw data sources to the platform, select the appropriate template which provides a set of predictions, then let the platform do all the hard work. The startup will automate data preparation and feature engineering, algorithm selection, and many of the time-consuming tasks in building models. The best part, the platform is built for the non-data scientist analyst. That’s a good thing because the addressable market for the BI analyst is significantly larger than that of the data scientist.


  • Company:
  • Founded: 2016
  • HQ: Tel Aviv
  • # of Employees: 35
  • Raised: $15M
  • Founders: Zohar Bronfman (CEO) and Noam Brezis (CTO)
  • Product: AutoML. AI platform that automates data preparation, feature engineering, and predictive analytics for the business analyst