Forecast how each ferment will finish.
NanoFlux calibrates a fermentation model to your own past runs, then forecasts time to finish and final gravity from the readings you already log. No new sensors.
Ferments are judged on a few manual readings a day.
Wineries, breweries and distilleries make in-run decisions from sparse, delayed measurements.
Readings are sparse
Gravity or Brix and temperature, logged by hand. They show where a batch is, not where it is heading.
Problems surface late
A slow or stuck ferment is often recognised after the fact, when tank time or a batch is already lost.
Digital twins don’t fit
Industrial twins need curated data, expert services and long validation, so most producers never use them.
So in-run calls still rest on rule of thumb, and the cost shows up as tank time and lost batches.
Three steps, on the data you already record.
A mechanistic yeast fermentation model, calibrated to your own history rather than a generic curve.
Calibrate on your history
We fit the model to your own past runs: gravity or Brix and temperature, plus yeast and pitch rate where you log them.
Forecast the finish
Each new reading updates the forecast of time to finish and final gravity for the batch in the tank.
Flag early, you decide
When a ferment drifts from its expected path, you see it while there is still time to act. Your team makes the call.
Built first for the cellar, then for scale-up.
Wineries, breweries and distilleries first. Then scale-up support for precision-fermentation facilities, where a failed pilot run is expensive.
A good fit if
- 01You already log gravity or Brix and temperature, by hand or in software.
- 02You want to know sooner when each batch will finish, and which ones are drifting.
- 03You would rather not install new sensors or start an integration project.
- 04You can share logs from completed ferments for a free benchmark.
Nothing in your process changes unless you decide it.
Early, and specific about it.
Pre-revenue. This is what works today, and what we test next.
Proof of concept in simulation
A working simulation engine with a parameter-calibration module, tested on synthetic data. TRL 3: no validation on real process data yet.
First real-data benchmark
Forecasts from the first 24 to 48 hours of readings on producers’ completed ferments, compared with their rule of thumb and naive extrapolation. Target: TRL 4.
In the cellar, then scale-up
Forecasts during the run for wineries, breweries and distilleries. Then scale-up support for precision-fermentation facilities.
We will report the benchmark result, positive or negative.
See how early your past ferments could have been called.
Free for producers. Retrospective: no access to your systems.
You share
Logs from completed ferments: gravity or Brix, temperature and the yeast used.
You get
A two-page report on how early each finish, and each slow ferment, could have been called.
What we compare
Forecasts from the first 24 to 48 hours of readings, against your rule of thumb and naive extrapolation.
Your data
Your data and results stay yours. Anonymised aggregates only with your written consent. Nothing changes in your process.
or write to hello@nanofluxbio.com
Two biotechnologists who studied together.
One builds the model; the other takes it to producers.
Alberto Fortes García
BSc Biotechnology, Universitat de València. MBA, EAE Business School and IL3-Universitat de Barcelona. Consultant in R&D and innovation funding at FI Group, then B2B sales. Co-founder of DoggyKit, winner of Hackathon Agrotech 2022.
Miguel Ángel Tirado
Pharmaceutical engineer. BSc Biotechnology, Universitat de València. MSc Pharmaceutical Design and Engineering, DTU. Founder of Youpaws, a probiotic pet-health startup. Builds the modelling engine and the calibration code.
DKK 50,000 Early Innovators Grant from the BioInnovation Institute, awarded through DTU Skylab in March 2025. Stage 1 of the DTU Skylab Startup Programme.
Send us your past ferments.
Producers, facilities and investors: write to us.