Pathway, founded in 2020, is an AI company developing what it describes as the world's first post-transformer frontier model. The company's core technical proposition is solving the problem of continual learning for enterprise applications, allowing models to learn from new data without retraining from scratch. This is achieved through its BDH architecture, which is designed to outperform traditional Transformer models and provide greater transparency into model operations.
The company's product suite combines this foundational model with what it states is the fastest data processing engine available on the market. The technology stack is targeted at enterprise AI use cases requiring contextualized, real-time intelligence from dynamic data streams.
Pathway's enterprise clients include NATO, the French postal service La Poste, and Formula 1 racing teams, indicating applications across defense, logistics, and motorsport. The company maintains offices in Palo Alto, California; Paris, France; and Wroclaw, Poland.
The technical team includes figures with notable credentials in AI research, such as a founder who is a complexity scientist and a CTO who worked with Nobel laureate Geoff Hinton. The company is also backed by a co-author of the Transformer architecture.
