It started as a weekend project. Precica's founder, Ajay, was running tsunami and storm-surge simulations on an HPC cluster — each run producing hundreds of gigabytes to terabytes of data that had to be post-processed immediately, then deleted, since there was nowhere to keep it all. So he built a way to compress the output, download just that, and query it directly in tools like ParaView — seeing the full simulation without ever downloading the full dataset.

The idea generalised fast, into a bigger question: could data stay useful, searchable, and forecastable without ever being unpacked? Four years later, that question is a company — spun out of the University of Manchester, backed by Northern Gritstone and Deeptech Labs, and already running pilots with industrial partners.

“We didn't set out to build a better compression algorithm. We set out to remove the tradeoff between storing data and using it.”