Cortica

Inspired by the brain, not by the amount of data

Cortica was founded in 2007 at the Technion by Prof. Yehoshua (Joshua) Zeevi and two of his doctoral students, Igal Raichelgauz and Karina Odinaev. Instead of relying on supervised learning, which requires huge amounts of manually labeled data, the company built a system based on principles from the way the brain's visual cortex processes visual patterns, which is where the company's name comes from. The internal mechanism, which the company calls Signatures, produces a compressed internal representation of visual and audio patterns, from which an Adaptive Architecture is built that recognizes context in real time.

Not a single-product company, but a parent of a portfolio

According to the company's own website, Cortica has invested more than $250 million over almost two decades in developing the technology, and has registered more than 300 patents. Instead of selling a single product, it built a portfolio of independent subsidiaries that apply its core technology in different fields: AutoBrains (originally Cartica AI) in autonomous vehicles, Corsight AI in face recognition, and other companies in aviation security and medicine. AutoBrains and Corsight AI have their own entries, and each of them has raised funding and developed its own business strategy.

The business model: spinning an application off into its own company

Cortica's model illustrates an approach that is becoming more common in the AI world: a core technology company does not sell directly to end customers, but spins off a specific application into a separate company with its own management team, funding and brand identity. For customers in the various industries (automotive, security, aviation, medicine), this means they work with the relevant subsidiary, but benefit indirectly from Cortica's long experience and investment in the core technology.

Rare longevity in a fast-changing industry

Unlike the Transformer-based approaches that have become dominant in the AI industry since the late 2010s, Cortica's technology originally developed in an independent direction, inspired by biological principles. Almost two decades of activity, starting in years when the term "artificial intelligence" itself was less common in public discussion, make it one of the interesting reference points for comparing different research approaches that developed in parallel in the industry.