How a Czech physicist helps brewers save beer with algorithms

The Czech startup YeastMaster, founded by physicist Ondřej Sghanel, helps breweries monitor the health of their yeast using machine learning — and in doing so saves tens of thousands of litres of beer a year that are currently lost to poorly managed fermentation.

By Sghanel's own calculations, around 3.9 million half-litre bottles of beer are poured away every year in Czechia because of yeast-related problems — enough, lined up end to end, to stretch from Prague through Brno all the way to Olomouc. That's precisely the problem his team set out to solve.

Ondřej Sghanel is now 42, and by training he's a specialist in electron microscopy and particle physics. The idea for the business came to him during an internship at an American brewery, where he noticed that even small breweries tracked fermentation quality in detail and squeezed out every percentage point of yield, whereas Czech brewers more often rely on experience and intuition rather than precise measurement.

The problem, the entrepreneur says, often lies not in the technology or the recipe but specifically in the state of the yeast — it determines how fermentation proceeds, as well as the beer's taste and stability. If a brewery fails to notice in time that its yeast is in poor shape, it can lose an entire batch of product along with the raw materials, the time and the customers involved.

That's how YeastMaster was born — a tool that lets brewers quickly assess the concentration and viability of their yeast without needing to run their own lab. The brewer takes a sample, stains it with a special dye (which colours only dead cells), takes photos under a microscope and uploads them to the program. Machine-learning algorithms automatically identify and count the yeast cells, producing a quality report and helping diagnose a common problem — when the beer "won't ferment" and the cause isn't clear. According to Sghanel, this is often what turns out to be the "killer" of small breweries.

The service also lets brewers calculate in advance the optimal amount of yeast to pitch for a brew, which can shorten fermentation by several days and prevent whole batches of beer from spoiling due to problems detected too late.

The project started in 2022, when Sghanel teamed up with Martin Čičala, and the two of them spent two years developing the technology — manually labelling thousands of images of live and dead yeast cells in the evenings to train the algorithm. Stanislav Tomanek later joined the team, the company was registered in Brno, and a pilot launch ran at several microbreweries in Brno. Since the start of last year, YeastMaster has officially offered its solution on the Czech market.