Jemba

Jemba is developing a B2B industrial machine-learning platform designed to turn existing factory data into actionable recommendations without requiring customers to employ data scientists. The platform is designed around the reality of existing industrial data infrastructure—including historians, PLCs, sensors and quality systems—rather than requiring manufacturers to deploy an entirely new instrumentation layer. Jemba argues that its technological differentiation comes from productizing the data-science workflow so that plant and process managers can operate it themselves.

Manufacturing companies and industrial groups seeking to improve production efficiency, maintenance, quality and energy performance using existing operational data. Jemba particularly targets the large number of individual factories for which conventional industrial-AI projects are economically impractical. Its thesis is that large manufacturers themselves operate many smaller plants that individually cannot justify six-figure AI deployments and dedicated analytics teams

Jemba identifies two principal categories of alternatives: 1. Vertical industrial-AI specialists building first-principles models and digital twins for particular processes. 2. Large industrial software/platform providers offering unified suites and customized deployments to large enterprises. Jemba argues that both approaches disproportionately serve the largest and best-funded plants, leaving a substantial portion of manufacturing capacity underserved.

Jemba's thesis is that manufacturers already possess much of the data infrastructure required for industrial machine learning. Years of investment in PLCs, historians, sensors and quality systems mean the remaining challenge is extracting useful decisions from that data economically and at scale. The company's technical foundation appears particularly strong in time-series analysis and unsupervised anomaly detection. Chief Data Scientist and apparent co-founder Alexandre Bois completed a PhD in applied mathematics at ENS Paris-Saclay and subsequently worked as a data scientist at TeepTrak. His profile describes his work as turning hundreds of raw process variables into anomaly detection and process optimization systems usable by industrial teams without specialist data-science expertise.

Separate an R&D-intensive industrial-AI product from TeepTrak's established monitoring business and build it as an independent software company. Jemba explicitly says an established monitoring company and an R&D-heavy venture have different “risk profiles, different capital needs and different horizons,” and that attempting to fund both from the same budget would disadvantage both businesses. The initial commercial proposition is industrial ML that can be deployed against existing plant data and operated by process engineers rather than data scientists.

B2B SaaS / enterprise industrial software. Alexandre Bois explicitly describes the venture as a “B2B SaaS Machine Learning platform for manufacturers.”