Rel Int provides an AI platform based on relational intelligence (RI), which breaks down global abstractions into explicit local interaction graphs to deliver transparent, step‑by‑step reasoning for each prediction.
Funding
Funding not disclosed
Founders
Product
Problem
Complex data environments often require AI models that are both highly expressive and transparent, yet existing solutions are either opaque black boxes or limited to classical hardware, making it difficult to trust and scale analyses across diverse scientific and industrial domains.
Solution
Rel Int delivers an AI platform built around relational intelligence (RI), a paradigm that decomposes global abstractions into local interactions, enabling distributed structure generation that is inherently explainable. The platform runs efficiently on any classical processor and can be mapped onto quantum hardware, allowing users to leverage quantum speedups without sacrificing interpretability. By exposing the underlying relational graph of a model, Rel Int provides clear, step‑by‑step reasoning for each prediction, facilitating validation and debugging. Open‑source toolkits such as DisCoPy and lambeq support developers in constructing and visualizing string‑diagram representations, bridging theoretical frameworks from physics, linguistics, and cognitive science to practical AI applications.
Target Audience
Primary customers are research labs, enterprises, and developers in fields such as quantum computing, natural language processing, and scientific modeling that require scalable, interpretable AI solutions.
Features
- Relational intelligence engine that transforms high‑level abstractions into explicit local interaction graphs for transparent inference
- Dual‑runtime capability that scales from standard CPUs/GPUs to quantum processors with minimal code changes
- Open‑source Python libraries (DisCoPy, lambeq) for building, visualizing, and executing string‑diagram based models
- Built‑in support for categorical quantum mechanics and ZX‑calculus, enabling direct implementation of quantum algorithms
- Automated distributed structure generation that exploits parallelism across classical and quantum resources
- Comprehensive explainability interface that traces predictions back through the relational graph for auditability