The predictive analytics platform built for the people who own the number - not only the ones who own the notebook.
For a decade, "we should use predictive analytics" ran into the same wall: it needed a data scientist, a clean pipeline, and a quarter of someone's time. Pecan AI's wager is that the wall was never the math. It was the gatekeeping.
Pecan is a predictive analytics platform that connects to a company's data warehouse and automatically builds, trains and deploys machine-learning models - for customer churn, lifetime value, demand forecasting, lead scoring and campaign return. The pitch is deliberately unglamorous: give the analyst who owns the retention number the ability to model it, without waiting in a data-science queue.
Founded in 2018 by Zohar Bronfman and Noam Brezis, two computational-neuroscience PhDs from Tel Aviv University, the company grew out of a prototype the pair built for a university data-science competition. Both are alumni of Unit 8200, Israel's elite signals-intelligence corps - a background that shows up in the platform's security-first posture.
In January 2024 Pecan named its approach "Predictive GenAI." The idea is a division of labor between two kinds of AI: a large language model helps a user frame the business problem in plain language through features called Predictive Chat and Predictive Notebook, and proven machine-learning techniques do the actual predicting. The chatbot sets the table; the model does the work.
It is a distinction worth dwelling on, because a lot of "no-code AI" is a demo that never survives contact with production data. Pecan's bet is that pairing conversational setup with real supervised learning is what makes the workflow durable enough for enterprises to trust with a forecast.
Pecan sells to enterprise and mid-market business and data teams across retail, e-commerce, gaming, consumer goods, financial services, telecom and healthcare. The common thread is a team sitting on customer or operational data that wants predictions it can act on - not another dashboard of what already happened.
Named customers span the beverage brand Hydrant, internet provider Clearwave Fiber, credit-repair firm The Credit Pros, retailer ShopTJC, mobile-game studio PlaySimple and European retailer ALTHERR. Clearwave built a churn model in four weeks and reported up to 20x lower churn in its highest-risk segment; The Credit Pros cut cancellations 25% within 30 days of intervention.
Two problems, really. The first is access: predictive modeling has historically required scarce, expensive data-science talent, so most business questions never got modeled at all. The second is speed: even when a team had the talent, a single model could take months, by which point the churning customer had already left.
Pecan attacks both by automating feature engineering, model training and deployment, and by letting a non-specialist describe the goal in words. The payoff its customers describe is a shift from post-mortems on lost accounts to interventions that arrive in time.
"Pecan makes AI-powered predictions accessible to all data and business teams by automating model creation and training."
The core cloud/SaaS product. Connects to a data warehouse and automatically builds, trains and deploys models for churn, lifetime value, lead scoring and campaign ROI.
A category pairing LLMs with AutoML. Predictive Chat and Predictive Notebook let users describe a problem in natural language to kickstart model building.
An AI co-pilot that guides business teams through defining and building predictive models - designed for organizations without dedicated data scientists.
A dedicated GenAI-powered demand-forecasting solution for medium-to-large enterprises, delivering accurate, explainable supply-chain forecasts with no coding.
The predictive/AutoML market is crowded - DataRobot, H2O.ai, Amazon SageMaker Canvas, Google Vertex AI, and lighter tools like Obviously.AI and Akkio all compete for it. Pecan's positioning is narrower and more opinionated: it is built for the business team, not the platform team.
Two design choices carry that stance. First, natural-language setup via Predictive GenAI, so the person defining the model is the person who understands the business goal. Second, explainability shipped with the output - because a forecast a planner cannot defend to their boss never leaves the slide deck.
Pecan is a B2B SaaS company. It sells recurring subscriptions to its platform, integrates with a customer's existing data warehouse and BI stack, and expands as teams add predictive use cases - churn, then lifetime value, then forecasting.
Public estimates put annual revenue in the neighborhood of $8M with a team in the high double digits, backed by roughly $127M in venture funding. The strategy is depth of adoption inside accounts rather than pure logo count.
"Predictive GenAI combines the strengths of large language models with proven machine learning."
| Round | Amount | Date | Lead / Investors |
|---|---|---|---|
| Seed & Series A | $15M | 2020 | Dell Technologies Capital, S Capital |
| Series B | $35M | May 2021 | GGV Capital, Vintage, Dell, S Capital, Mindset |
| Series C | $66M | Feb 2022 | Insight Partners, GV, GGV, S Capital, Dell, Mindset, Vintage |
Cumulative capital raised (USD, approximate)
Figures reflect cumulative disclosed funding through the 2022 Series C.
Holds two PhDs from Tel Aviv University - in computational neuroscience and in the philosophy of science - plus a BA in economics. An expert in computational psychology and data science, and a Unit 8200 alumnus.
PhD in computational neuroscience with an MS in cognitive psychology, plus a decade in software and data consulting and a cybersecurity background from his IDF service.
The company's expertise sits at an unusual intersection: cognitive science and applied machine learning, wrapped in an intelligence-corps discipline around data security. That combination shapes a product built to be understood by non-experts while still standing up to enterprise scrutiny.
Bronfman and Brezis found Pecan after a prototype wins attention at a Tel Aviv data-science competition.
Combined seed and Series A led by Dell Technologies Capital and S Capital.
GGV Capital leads a round to scale the predictive analytics platform.
Insight Partners leads with GV and GGV, bringing total funding to about $127M.
Predictive Chat and Predictive Notebook blend LLMs with automated machine learning.
Pecan ships a predictive-modeling co-pilot, launches DemandForecast.ai, and is named a Gartner Cool Vendor.
Pecan sits between two worlds that rarely meet. On one side are heavyweight data-science platforms built for specialists who write code and tune pipelines. On the other are business-intelligence tools that describe the past but do not predict the future. Pecan aims squarely at the gap: forward-looking predictions owned by the business.
Its 2025 moves - a predictive co-pilot and the supply-chain-focused DemandForecast.ai - read as a bet that the next wave of adoption comes from function-specific products aimed at planners, marketers and retention teams, rather than a single general-purpose tool. The Gartner Cool Vendor recognition in supply-chain technology points the same direction.
Product walkthroughs and founder interviews live on Pecan's channels and press coverage:
▶ Pecan AI on YouTube ▶ Predictive GenAI demos ▶ Zohar Bronfman interviewSources include Pecan AI press releases and company pages, BusinessWire, VentureBeat, SiliconANGLE, Calcalist/CTech, The Jerusalem Post, MarTech Series and Crunchbase. Financial and headcount figures are approximate and drawn from public reporting.