23+ Patents. One Mission.
The AI engine behind Tabrid.ai represents a decade of industrial AI research by Carnotfleet Korea — the deepest patent portfolio in thermal intelligence for the GCC.
Patent Portfolio Overview
23+ registered patents spanning four core technology domains. Each domain represents a distinct moat that competitors cannot replicate without years of research investment.
Registered AI & IoT Patents
Filed across Korea, UAE, Saudi Arabia, and under PCT international patent agreements — protecting Tabrid.ai's technology in every target market.
Thermal Prediction
7+Recurrent neural network architectures trained to forecast rack-level temperatures up to 30 minutes ahead, accounting for workload variability, ambient conditions, and cooling system lag.
Anomaly Detection
6+Unsupervised and semi-supervised models that identify deviating thermal signatures without requiring labeled training data — critical for novel failure modes that have never been seen before.
Energy Optimization
5+Reinforcement learning agents that optimize chiller and CRAH setpoints in real time, balancing thermal compliance objectives against energy cost minimization across complex HVAC topologies.
IoT Data Fusion
5+Methods for fusing heterogeneous sensor streams — temperature, humidity, airflow, power, vibration — into unified thermal state representations that enable higher-order inference than any single signal permits.
AI/ML Architecture
A production-grade machine learning pipeline designed for the latency, reliability, and data quality requirements of industrial thermal environments.
Stage 1
Data Ingestion
High-frequency telemetry from thousands of IoT sensors is ingested via MQTT and REST APIs, normalized to a canonical schema, and validated for quality before entering the processing pipeline.
Stage 2
Feature Engineering
Raw sensor readings are transformed into model-ready features: rolling statistics, lag features, ambient correction terms, and cross-sensor correlation signals derived from facility-specific topology graphs.
Stage 3
Model Inference
Edge-deployed inference models run on Tabrid.ai gateway hardware for sub-100ms response times. Cloud models handle longer-horizon forecasting, optimization, and root cause analysis workloads.
Stage 4
Continuous Learning
Production predictions are compared against ground truth outcomes. Detected drift triggers model retraining pipelines that update edge models with facility-specific adaptations while preserving the core trained weights.
Carnotfleet Korea
The AI engine behind Tabrid.ai was built by one of Korea's most prolific industrial AI research teams.
Founded in Seoul, South Korea
Carnotfleet Korea was established by researchers who specialized in industrial process optimization at KAIST (Korea Advanced Institute of Science and Technology). The company focuses exclusively on AI applications for thermal management, energy systems, and industrial IoT.
- Exclusive focus on industrial thermal AI
- R&D team with combined 120+ years of experience
- Active publication in peer-reviewed journals
- 23+ registered patents across four core domains
- Clients in manufacturing, logistics, and data centers across Asia
Partnership with Arabkor Group
The technology-to-market partnership between Carnotfleet Korea and Arabkor Group brings Korean industrial AI to the GCC. Arabkor Group contributes regional infrastructure, regulatory expertise, and distribution relationships built over decades of operations in Saudi Arabia and the UAE.
- Arabkor provides GCC market access and operator relationships
- Carnotfleet maintains full IP ownership and R&D roadmap control
- Joint engineering team adapts models for GCC ambient conditions
- Shared deployment infrastructure across Riyadh and Seoul offices
- Co-developed GCC-specific training datasets for model adaptation
Why Korean AI?
Korea has invested heavily in industrial AI research, producing a generation of engineers who understand both the mathematical foundations and the operational constraints of factory-floor deployment.
Academic Foundation
Korea ranks among the world's top five nations for AI research output per capita. KAIST, Seoul National University, and POSTECH produce industrial AI researchers who understand real-world deployment constraints alongside theoretical foundations.
R&D Investment
Carnotfleet Korea's R&D team publishes regularly in IEEE, ASHRAE, and ACM journals. Every product feature is grounded in peer-reviewed research — not proprietary heuristics that cannot be validated or reproduced.
Industrial Context
Korea's semiconductor fabrication, display manufacturing, and data center industries have created one of the world's most demanding proving grounds for thermal AI. The algorithms deployed in Korean fabs are now the same algorithms optimizing GCC data centers.
Peer-Reviewed Research
Published in IEEE & ASHRAE Journals
The core algorithms powering Tabrid.ai have been independently validated through peer-reviewed publication. No black boxes — the science is open.
See Tabrid.ai in Action
Schedule a personalised demo with our solutions team and discover how AI-powered thermal intelligence can transform your operations.
Request a DemoNo commitment required. Typical demo is 30 minutes.