Company
Structure grows on its own, and insight grows more precise.
The name
The name kernalic is a nod to the kernel in deep learning. In a convolutional neural network, the kernel acts as a feature extractor, pulling meaningful features out of raw data. kernalic means 'kernel-like' — a company that pulls meaningful structure and relationships out of scattered data. The name is also the name of our method: Kernel Ontology, in which a person sets the broad frame and the data itself organizes and generates the relationships, building up the network.
What we do
kernalic is a software company that builds structure into data. We are not tied to one industry: wherever data is entangled, structured or not, we define how they relate and turn that structure into a product. Of the many domains ahead, the one with products out today is intellectual property. SEPPi, for standard-essential patent determination, runs on this principle today, and IPSnap, which ties patent claims to real products for infringement analysis and goes on to prior art and grounds for invalidity, will be built on the same principle.
Where we're headed
Data is already piled up faster than anyone can read it. The problem is not the volume but that, inside the pile, nobody can see what connects to what. Our direction is to find the relationships hidden in that mess and build them into a structure. Once relationships are structured, AI can reason over that network with the evidence attached, and keeps raising its precision on its own. With every new organization and domain we meet, we will keep finding the structure inside it, defining a new kernel for it, and growing that out as a Kernel Ontology.