Pinakes

2,300 years ago, the Library of Alexandria accumulated more knowledge than any scholar could navigate. Callimachus created Pinakes: a system for organising knowledge so that it could be found and understood by humans.

​

It was a system built for human intelligence. Machine intelligence operates at an entirely different scale.

​

Computing intelligence is scaling by orders of magnitude faster than knowledge access. Millions of AIs need to access vast amounts of information at machine speed every second of every day, dwarfing the Alexandrian scholars’ problem both in scale and speed.

​

This problem will grow to unfathomable proportions. More AIs trawling through larger datasets ever faster, means that AI efficiency is increasingly determined not only by the speed of computation, but knowledge  navigation  and access.

​

All knowledge must therefore become machine-navigable. For this to happen, it must be continuously structured, traversed and assembled at the speed and cost of reasoning itself. All existing solutions are orders of magnitude more expensive than this target.

​

This is an emerging workload that requires novel  infrastructure to execute economically.

​

Pinakes is defining a new hardware category: an architecture where knowledge can be persistently stored, navigated and accessed as a native function of machine intelligence.

​

We will enable a world in which intelligence is no longer constrained by what can fit inside a model or be retrieved from a database. Knowledge will be accessible to everything. Millions of AIs can reason over a shared, continuously evolving representation of knowledge: connecting discoveries across disciplines as they emerge and building on what every intelligence before them has learned.

​

Our ambition is to make the world’s knowledge addressable at the scale of machine  intelligence.

​

​

hello@pinakes.co