The company's goal is to create a system that can perform semantical processing of programming code, e.g. code analysis, transformation, or synthesis, as well as operate a computational environment. The approach combines explicit knowledge representation in discrete structures with ML-assisted algorithms to manipulate those structures. The system will have the following core properties.
The company's goal is to create a system that can perform semantical processing of programming code, e.g. code analysis, transformation, or synthesis, as well as operate a computational environment. The approach combines explicit knowledge representation in discrete structures with ML-assisted algorithms to manipulate those structures. The system will have the following core properties.
The company's goal is to create a system that can perform semantical processing of programming code, e.g. code analysis, transformation, or synthesis, as well as operate a computational environment. The approach combines explicit knowledge representation in discrete structures with ML-assisted algorithms to manipulate those structures. The system will have the following core properties.
Interpretability
Thanks to using discrete structures like combinatorial cell complexes augmented with language and semantical data, a user will be able to follow the reasoning trajectory and examine the internal logic of any inference process. This will guarantee the system’s safety and reliability.
Interpretability
Thanks to using discrete structures like combinatorial cell complexes augmented with language and semantical data, a user will be able to follow the reasoning trajectory and examine the internal logic of any inference process. This will guarantee the system’s safety and reliability.
Interpretability
Thanks to using discrete structures like combinatorial cell complexes augmented with language and semantical data, a user will be able to follow the reasoning trajectory and examine the internal logic of any inference process. This will guarantee the system’s safety and reliability.
Efficient Extendability
Contrary to the conventional large language models approach, we explicitly represent latent knowledge. This way, incorporating new knowledge at later stages of system development will be as inexpensive as at the initial ones. This quality will make system updates faster and cheaper.
Efficient Extendability
Contrary to the conventional large language models approach, we explicitly represent latent knowledge. This way, incorporating new knowledge at later stages of system development will be as inexpensive as at the initial ones. This quality will make system updates faster and cheaper.
Efficient Extendability
Contrary to the conventional large language models approach, we explicitly represent latent knowledge. This way, incorporating new knowledge at later stages of system development will be as inexpensive as at the initial ones. This quality will make system updates faster and cheaper.
Interactivity
The system is designed to have an internalised notion of bounded rationality and non-omniscience, so it will be able to identify a lack of information and interact with external sources to achieve designated objectives.
Interactivity
The system is designed to have an internalised notion of bounded rationality and non-omniscience, so it will be able to identify a lack of information and interact with external sources to achieve designated objectives.
Interactivity
The system is designed to have an internalised notion of bounded rationality and non-omniscience, so it will be able to identify a lack of information and interact with external sources to achieve designated objectives.
Hierarchical Contexts
The system will adjust its behaviour depending on the user-referred context: programming language, project, user, or current session. This way, the system will choose between abstract domain-independent knowledge and narrow domain-specific knowledge accordingly.
Hierarchical Contexts
The system will adjust its behaviour depending on the user-referred context: programming language, project, user, or current session. This way, the system will choose between abstract domain-independent knowledge and narrow domain-specific knowledge accordingly.
Hierarchical Contexts
The system will adjust its behaviour depending on the user-referred context: programming language, project, user, or current session. This way, the system will choose between abstract domain-independent knowledge and narrow domain-specific knowledge accordingly.