This article introduces Fast-Cat, an autonomous self-driving catalysis laboratory that enables efficient exploration of reaction parameter spaces and Pareto-front mapping for high-temperature, high-pressure gas-liquid reactions. Fast-Cat autonomously benchmarks ligands and evaluates catalyst performance with minimal human intervention, using a modular flow chemistry platform and machine learning (ML) techniques. The system rapidly identifies optimal experimental conditions for hydroformylation of 1-octene using various phosphorus-based ligands, demonstrating its ability to map the ligand-regioselectivity-yield relationship in just 5 days with no human involvement. Fast-Cat generates high-quality experimental data to build a digital twin of the reaction system, enabling virtual analysis of reaction mechanisms and process parameters. The system's scalability is validated by transferring knowledge from a flow reactor to a batch reactor, showing consistent results. Fast-Cat's ML brain uses Bayesian optimization to select experimental conditions that improve reaction yield and regioselectivity, with Shapley analysis revealing the impact of key process parameters on reaction outcomes. The system's autonomous capabilities significantly enhance catalyst discovery and development by accelerating reaction space exploration and reducing experimental costs. Fast-Cat's modular design allows for future integration with robotic chemical workstations and other automation tools, expanding the range of materials available for reaction space exploration. The study highlights the potential of autonomous systems in catalysis, demonstrating their ability to rapidly identify optimal ligands and reaction conditions for complex chemical transformations.This article introduces Fast-Cat, an autonomous self-driving catalysis laboratory that enables efficient exploration of reaction parameter spaces and Pareto-front mapping for high-temperature, high-pressure gas-liquid reactions. Fast-Cat autonomously benchmarks ligands and evaluates catalyst performance with minimal human intervention, using a modular flow chemistry platform and machine learning (ML) techniques. The system rapidly identifies optimal experimental conditions for hydroformylation of 1-octene using various phosphorus-based ligands, demonstrating its ability to map the ligand-regioselectivity-yield relationship in just 5 days with no human involvement. Fast-Cat generates high-quality experimental data to build a digital twin of the reaction system, enabling virtual analysis of reaction mechanisms and process parameters. The system's scalability is validated by transferring knowledge from a flow reactor to a batch reactor, showing consistent results. Fast-Cat's ML brain uses Bayesian optimization to select experimental conditions that improve reaction yield and regioselectivity, with Shapley analysis revealing the impact of key process parameters on reaction outcomes. The system's autonomous capabilities significantly enhance catalyst discovery and development by accelerating reaction space exploration and reducing experimental costs. Fast-Cat's modular design allows for future integration with robotic chemical workstations and other automation tools, expanding the range of materials available for reaction space exploration. The study highlights the potential of autonomous systems in catalysis, demonstrating their ability to rapidly identify optimal ligands and reaction conditions for complex chemical transformations.