Pecan

When building a prediction model takes weeks

Building a prediction model, such as a customer churn forecast or a demand forecast, usually requires a skilled data scientist to choose variables, train a model and calibrate it. That process can take weeks for each new business question. Pecan was founded in 2018 by Zohar Bronfman (CEO, who holds two doctorates from Tel Aviv University, in computational neuroscience and in the philosophy of science) and Noam Brezis (PhD in computational neuroscience). It developed an automated predictive analytics platform that cuts out much of this manual work and lets analysts and data teams without a data science background build prediction models themselves.

From the business question to the answer, without a data scientist

The user defines the business question they want to predict, for example which customers are likely to leave, what demand for a certain product will be next month, or which transactions are suspected of fraud. The system automatically handles the technical steps that usually require dedicated expertise: choosing variables, training models and evaluating their accuracy. The approach is meant to make AI-based prediction accessible to organizations that do not have a large team of data scientists.

Leadership: Trevor Healy as executive chairman, and Tomer Miron as CTO

To strengthen its leadership, Pecan also brought in Trevor Healy as executive chairman, a move that signals the company's ambition to keep expanding in the enterprise analytics market. Tomer Miron, a senior engineer with more than 20 years of experience at companies such as eBay and Google, serves as the company's CTO and SVP R&D, according to the official company page. Pecan positions itself as a bridge between the technically complex world of predictive AI and business and analytics teams that want quick answers without training in programming or data science.

AutoML: a broad trend in the AI industry

The term AutoML itself reflects a broad trend in the AI industry. As more organizations want to use data-based prediction, it becomes less feasible to employ a dedicated data scientist for every single business question. Not every organization has the budget or enough need for such a full-time role. Platforms like Pecan try to bridge this gap by packaging that professional knowledge into a guided workflow, so that a business analyst, and not only a qualified data scientist, can run the complex steps themselves.

2026: a first-of-its-kind "predictive AI agent"

In January 2026, according to an official announcement, Pecan launched what it calls the "World's First Predictive AI Agent": a tool that lets a business team feed in raw data, ask a question in plain language, for example which customers are likely to churn, how demand will change, or which deals are close to closing, and get a statistically validated, production-grade prediction within minutes, without involving a team of data scientists. In the official announcement, Pecan CEO Zohar Bronfman said that for years businesses have invested billions in data platforms, dashboards and analytics, and yet the most critical decisions are still made too late. The tool had already been tested privately by existing customers, including the company Bubble.