Introducing PortexAI

There exists a paradox in the current age of AI.

While data is intuitively the world's most valuable commodity, it remains the only commodity for which there is no public venue for exchange. Instead, data today is siloed, secretive, and unstructured, even within major AI labs where post-training teams have superficial knowledge of their model's training data. We believe the way this contradiction is resolved will determine the course of AI innovation, especially in pre-training.

The Portex founding team traces its roots to the R&D arm of Coin Metrics, a pioneer in the field of blockchain analytics and data. Working within Coin Metrics’ experimental Labs unit, we bonded over a profound curiosity to decipher blockchains with empirical data, building innovative metrics to measure new industries. We now want to apply this knowledge to AI, and help the industry better reason about pre-training datasets and performance using better metrics.

As for our name, Portex is a reflection of our shared interests in history and the origins of modern cryptography. The Portex was a British electromechanical cipher machine used by secret services in the UK during the late 1940s and 1950s. It was similar to, but more advanced than, the infamous German Enigma, and pays homage to the early days of cryptography as an emerging field. Tokenized data in raw format can resemble the scrambled outputs of a cipher machine, hence our efforts to decipher this emerging space. 

Stay tuned 

If you are interested in accompanying us on our journey and research, make sure to follow @portexai on Twitter/X and subscribe to our Research below. 

The Portex founding team is Lucas Nuzzi (X, Linkedin), Kyle Waters (X, Linkedin), Tadhg Looram (Linkedin)