We are taking HyperDimensional Computing (HDC) to the next level, developing a novel enabling AI technology.
Our goal is to leapfrog current language-based solutions to Artificial General Intelligence by building an experience-based intelligence, capable of performing complex cognitive tasks.
The AI market got $1.3 trillion in investment in the last 5 years, under the premise that, throwing more data and more computational resources, the technology behind ChatGPT will produce AGI.
However, current AGI front-runners are unable to develop AGI due to their reliance in text. This could lead to the loss of $40 trillion from Nasdaq if they do not change course.
Our company is committed to develop a four point plan to avoid the creation and burst of an AI bubble:
Developing AGI
Standardizing It
Proving Its Utility
Deploying It Into The Economy
Our science team focus on developing the next generation of AI with the goal of achieving AGI. They have a plan built over a tested technology called HyperDimensional Computing.
Our technology team focus on transforming our science into production ready code. They develop our main product:
The Kogniment API.
Our applicability team focus on building Proofs of Concept of how to use our API in the real world. Their goal is to show to potential customers the capabilities of our technology.
Our sales team focus on actively deploying our technology into the economy, ensuring a proper adoption of the novel capabilities AGI provides.
HyperDimensional Computing is a novel representation-based computational framework that has been able to solve complex cognitive tasks like Raven’s Progressive Matrices and the Tower of Hanoi logic game that ChatGPT was unable to solve.
Current advanced AI solutions are based on language-based intelligence, missing that link with the real world. Our AI solutions are based on experience-based intelligence, where the real world shapes its knowledge.
Our solutions are founded over a representation based approach. This makes our solutions robust to weird quirks like adversarial attacks or hallucinations. They are also energy-efficient due to reduced training requirements. Finally, they are not only transparent and explainable, but also human-interpretable.
We provide an API to manage our AGI technology. It is a Python library that clients can use to develop their finalist solutions.
Our API provides tools to define an embodiment for our technology by defining connections to data sources (i.e. a camera, a dataset, etc…) and to actuators (i.e. a motor, a classification label, etc…).
Our API provides tools to control training and inference, and to explore the learnt knowledge representation. The client is the one responsible for the adequate training of the model.
We are a committed team of 6 people, lead by our founder and lead scientist. Both his expertise in AI research and his strong mathematical background gave him a great knowledge about the state of the AI field and the limitations of current AI solutions. With this knowledge, he has taken the step to launch his own startup to develop his vision of an Artificial General Intelligence, one capable of true reasoning.