Pecan AI, which is developing a predictive analytics platform for enterprises, raises $66M

“Big data analytics” is perhaps the currency of the pandemic. The term describes the process of uncovering trends, patterns, and correlations in large amounts of data to help make decisions. Big data analytics is being used by a growing number of organizations to anticipate trends across industries — according to a 2020 NewVantage Partners survey, 64.2% of businesses claim that they’re using data to drive innovation.

The promised advantages of big data analytics include faster innovation cycles, improved business efficiencies, higher productivity, and more effective R&D. Using big data, Netflix reportedly saves $1 billion per year on customer retention. But implementing big data infrastructure within an organization is often fraught with challenges, including poor data quality and a lack of expertise.

With the big data analytics market set to grow to as high as $100 billion in value by 2027, vendors — looking to get in on the ground floor — are offering new services designed to abstract away the complexities of working with big data. Noogata, Imply, and Unsupervised are among the companies offering predictive modeling, anomaly detection, and other tools that ingest large amounts of data and process it with AI to identify insights. Pecan AI also falls into this category — the company works with customers to automate data cleansing and engineering to create machine learning-based predictive algorithms. Marking its second round in the last 12 months, Pecan today announced that it raised $66 million in a series C round led by Insight Partners with participation from S-Capital, GGV, Dell Technologies Capital, Mindset Ventures, and Vintage Investment Partners, bringing its total capital raised to $117.5 million.

Predictive analytics

Tel Aviv, Israel-based Pecan was founded in 2016 by Noam Brezis and Zohar Bronfman. Bronfman has Ph.D.s in the history and philosophy of science and technology and computational cognitive neuroscience from Tel Aviv University. Prior to joining Pecan, Brezis was a senior data consultant at Madiera Data Solutions, a firm that’s worked with customers including Intel on data-focused products and services.

With Pecan, users can connect to a variety of databases and use a drag-and-drop, no-code interface to create machine learning datasets. The platform can enrich the datasets using outside data and automatically perform feature engineering, the process of selecting the most relevant variables from data to represent a business problem to a statistical model.

“The scarcity of highly-skilled analytical talent is a challenge that every industry is facing. Every year, we generate significantly more data than the year before — but data isn’t useful on its own without the proper resources to analyze and visualize it,” Bronfman told VentureBeat via email. “Data science can drive profound business impact, but it has to be implemented by skilled data scientists and data engineers. [M]ost companies don’t have sufficient data science talent and resources to analyze and optimize their business performance. Pecan helps companies leverage data science without requiring data scientists in the loop.”

Pecan trains and optimizes models over time, prioritizing features as they change in importance and showing the changes in a live monitoring dashboard. As the company explains on its website: “Matching the nature of your data, its size, and the predictions you require, Pecan constructs a large number of deep neural networks. After a … set of recursive competitions between multiple networks, only one fully trained neural network is left — refined for optimal performance and accuracy.”

The idea is to let analysts and business stakeholders obtain actionable insights and see outputs in a matter of days after adding their raw data. Pecan supports use cases that include demand forecasting, conversion, lifetime value, next best offer, VIP customers, upsell and cross-sell, churn and retention, and sales analytics.

“Because Pecan focuses on data analysts, our AI-automation platform is use case focused, which greatly reduces the complexity and statistical knowledge required from its users,” Bronfman added. “Unlike some of our competitors, Pecan handles everything from data prep to modeling and monitoring production with a drag-and-drop and low-code SQL-based UI. The data prep and feature selection/engineering components are critical. Getting data into proper form for AI models can take weeks or months when working with data scientists and data engineers, but with some help from the Pecan team and from our platform, this can be automated with minimal effort.”

Implementation challenges

While big data analytics has its benefits, enterprises frequently encounter hurdles while adopting it. As Harvard Business Review (HBR) notes, predictive analytics is especially easy to get wrong — particularly if companies don’t prepare staff to manage the machine learning integrations and jump into modeling before establishing a path to operational deployment.

“Predictive analytics isn’t a technology you simply buy and plug in. It’s an organizational paradigm that must bridge the quant/business culture gap by way of a collaborative process guided jointly by strategic, operational, and analytical stakeholders,” HBR’s Eric Siegel writes.

For its part, 90-employee Pecan — which claims that recurring annual revenue tripled over the past year — points to success stories like Johnson & Johnson, which used its platform during the pandemic to help predict consumer behavior and buying patterns as well as supply chain resilience.

“Pecan has several dozens of customers in mid-market and enterprise segments, spanning fintech, insurance, retails, consumer packaged goods, mobile apps, and consumer services. The funding [from this latest round] will be used to expand our operations in Israel and the U.S.,” Bronfman said. “We expect to double our headcount over the next 12 months.”