Dependable
and
High-performance
AI Technologies



At Depintel, we believe that artificial intelligence and machine learning should not only help people solve problems effectively and efficiently, but also provide solutions that are interpretable and trustworthy.

We provide advanced software and professional services for high-performance AI applications that require a high degree of reliability. We specialise in research and development of novel AI techniques that produce auditable predictive models that are explainable and trustworthy.

Our Technologies & Services

AI Solutions

Silas: High-performance structured data analytics with eXplainable AI; DeepDebugger: Visualisation and explanation of deep learning training procedures; ReaLLM: enhancing Large Language Models with reliable reasoning.

Formal Reasoning

Rigorous modelling, simulation, reasoning and verification technologies for software and hardware systems provided by Process Analysis Toolkit (PAT).

Sports Analytics

Training plans and in-game tactics analysis through deep learning, advanced video processing, and probabilistic reasoning. Match outcome prediction and odds calculation used by bookmakers around the world.

Blockchain Analysis

Ensuring the correctness and security of smart contracts and blockchain protocols using PAT. Analysing illegal blockchain transactions and user behaviours using customised AI pipelines.

News Room

Depintel is working closely with univerisites around the world to host the International Sports Analytics Conference and Exhibition (ISACE) series 2025 in Shanghai, China, on in September 2025.

Our eXplainable AI (XAI) tool DeepDebugger is released to help the user visualise and understand deep learning models and improve their performance.

Silas v0.8.7 is released with better machine learning performance, higher computational efficiency, eXplainable AI capabilities, and more.

N-PAT is released as an extension of the Process Analysis Toolkit that can perform (probabilistic) model checking on nested and hierarchical systems with parallelism.

Contact us