Farmless Proteins
We use energy to make simple molecules from air, water and minerals. Then nature takes over. Fermentation turns them into nutrient-dense postbiotic protein, using a fraction of the land.
Full stack hard,
soft and wetware
Our fermentation runs continuously at extraordinary speed. Getting there took plenty of failures, late nights and a full-stack approach: biology, hardware and software designed as one system.
Conventional fermentation equipment was never built to grow this fast or run continuously. So we built our own controls, pumps, cooling systems and more.
The organism now grows so fast that biology is no longer the bottleneck. The limits are engineering: how much oxygen we can deliver and how quickly we can remove heat.
Conveniently, engineers enjoy those problems. Check out our open roles.
BrewControl
A continuous fermentation runs for weeks. The software controlling it has to do the same.
BrewControl is our control system, written from scratch in Rust. It reads live data, controls pumps and valves accordingly and keeps the process where we want it.
We control everything through mixes of PIDs, cascades and model-predictive controllers.
Instead of the usual tower of PLC, SCADA and historian software, we built one system we own from top to bottom that allows for modern controls.
We regularly build our own stacks or interfaces like custom PCBs, firmware or an MQTT system. Industrial equipment connects through S7, Modbus, EtherCAT and serial
New control algorithms can then run as sandboxed WebAssembly modules, letting us improve the process without making the core runtime fragile.
BrewBase
A fermentation run is more than a time-series graph.
Recipes, ingredients, set-points, samples, assays, harvests and sensory scores all live in BrewBase: one strongly typed Postgres data model built around the physical process.
We version process development like software. Every recipe and set-point change has an exact history, giving us traceability from a finished protein all the way back to the fermentation and ingredients that made it.
More importantly, R&D becomes queryable.
“Did running warmer improve flavour?”
“Which temperature gave the highest yield?”
“Why did this batch behave differently?”
Those are now questions for BrewBase or our Slack bot, rather than detective work across four spreadsheets.
Manifold
We're building the plant twice. First in software. Then in steel.
Manifold is our end-to-end model of the production process: fermentation, harvesting, equipment, mass flows, energy and economics.
Oxygen transfer and cooling define our fermentation boundary. Working volume, dilution rate and titre determine downstream capacity. Equipment sizing turns into energy demand, capex and ultimately cost per kilogram.
Manifold then turns the process into a real plant: 3D layouts, piping, PFDs and P&IDs, all tied to the tags, revisions and equipment data.
Supplier quotes are also managed here, linked directly to the equipment they price, so we can test real specifications in different scenarios.
We build the process models ourselves, calibrate them against pilot data and update them with real measurements.
This gives us a fast, rigorous way to iterate decisions, with just in time production.
The system is the company.
Everything here is coupled. Good biology cannot make up for unreliable equipment, good hardware cannot rescue poor control, and machine learning cannot repair a bad valve.
So we are a vertical integrator. We design the process, the controls, the software and the plant ourselves, and buy what is already a commodity: energy and steel. The integration is the innovation.
Each one feeds the next. Every run trains the models, and the simulations of one generation design the one after it.