Lemnisca builds software to predict how fermentation behaves at factory scale
The Bengaluru company is building AI software that models microbial fermentation, so manufacturers can scale a process without discovering its problems in production.
Microbes can be engineered to produce almost anything a plant or an animal makes: proteins, flavours, fuels, materials. Getting microbes to work in a laboratory flask is usually the easy part. The harder step is making them behave the same way in a tank that holds tens of thousands of litres.
The reason is that scale changes the physics. Oxygen reaches cells differently, mixing is uneven, heat builds where it did not before, and the organism responds to all of it. A process tuned over months in a flask can behave unrecognisably in production, and the failure usually appears only after the plant has been built.
was founded in Bengaluru in 2025 to predict that behaviour in advance.
Its founders are Pushkar Pendse, the CEO, and Shilpa Nargund, the CTO, who had earlier worked at a German company in the same field. The problem they set out to address sits in the gap between the laboratory and the factory, which is where they say most of the industry's losses accumulate.
A digital twin for a fermentation tank
The product is what the industry calls a digital twin: a software model of a physical process, built to simulate how it will behave under conditions it has not yet been run in. Lemnisca describes its version as an AI companion for fermentation.
What the company says distinguishes its approach is that the model is science-aware, combining biology, reactor physics and data rather than pattern-matching on past runs alone. It pairs wet-lab experimentation with computational modelling, so the software is trained against real fermentation rather than only against historical datasets.
The commercial claims attached to it are specific: development timelines are cut by up to 50%, operating efficiency is improved by 25%, and there is faster adoption of renewable and waste-carbon feedstocks. These are the company's own projections; deployment data supporting them is not publicly available.
Nargund has framed the underlying problem as one of cost. Biotechnology has succeeded at making expensive molecules, insulin among them, at prices in the thousands of dollars per kilogram. The question she has posed is how the same methods reach products that have to sell for a few dollars a kilogram, which is where scale and predictability decide everything.
A pre-seed round, and a crowded global field
Lemnisca raised an undisclosed amount in a pre-seed round announced in November 2025, led by Theia Ventures with PointOne Capital and Dr Satakarni Makkapati participating. The money is for expanding its Bengaluru laboratory, developing the platform and running early pilots.
Theia Ventures, which has backed several early deep-technology companies working on industrial decarbonisation, has said biomanufacturing needs digital simulators capable of predicting how microbial processes will perform at industrial scale.
The company intends to work first with fermentation-led manufacturers already some way into their own scale-up, while building an internal product pipeline alongside. It has said it is inviting pilot collaborations with contract manufacturers, equipment makers and ingredient brands.
The startup's global competitors include Differential Bio in Germany and Invert Bio in the United States, which offer comparable services. Nargund has declined to name customers, citing confidentiality, and no pilot, contract or revenue figure is public. The company is still early-stage, and its platform is being validated through pilot work.
(This story has been researched and compiled using publicly available information.)