Josh Wong, Founder & CEO of ThinkLabs AI – Interview

Can you tell us more about the vision behind ThinkLabs AI and what inspired its creation?

The vision behind ThinkLabs is a reliable, sustainable, and affordable energy infrastructure powered by trustworthy AI. We understand that the grid remains at the center of the energy transition. To decarbonize we must electrify. To electrify we need the grid, and the grid really must modernize. We believe the intersection of electric power systems engineering, AI, and cloud computing is the solution.

How does ThinkLabs AI differentiate itself from other AI startups in the grid management sector?

The grid is complex, and so much so that AI in itself cannot learn about the complex power flows and operational processes that exist in the grid space. ThinkLabs combine the rich history and confidence of traditional power systems engineering with AI, as trustworthy physics-informed AI, for confidence in scaled, automated inferencing and decision support for critical infrastructure. It also takes more than technology, but an experienced team that understands the nuances of the grid and how utilities and regulators think. Our team comes from the electric power systems space with proven track record, including founder Josh Wong who has sold his previous company Opus One Solutions to GE, and stands at the intersection of engineering, AI, and cloud computing.

What is the ThinkLabs Copilot, and how does it enhance grid planning and operations?

ThinkLabs Copilot is a digital assistant that that understands the real world with proprietary physics-informed AI digital twins that provide a foundation model for engineering systems. It works with utility planners and operators, to model the grid into its “AI digital twin”, perform high speed and large scale analytics including in near real time, and make recommendations on grid operations, plans, and designs.

Can you explain what a physics-informed AI digital twin is and how it benefits grid reliability?

AI by itself can’t learn such a complex system as the grid with measurement data only. AI digital twins of the real world are trained by, work for, and work with engineering systems, hence “physics-informed”. Training is done using large amounts of synthetic data generated from engineering simulation. Traditional physics-only, impedance-based digital twins are deterministic and mathematically optimized, yet challenged by data quality, high computing power needed, and slow response time. Conversely, general AI techniques promise speed, yet sparse data, hallucinations, and “black box” effects concerns mission critical grid operations. A physics-informed AI digital twin offers transparent and trustworthy analytics, resilient and robust against bad data, fast response and action suitable for real time operations, preparedness with large pre-trained operating scenarios, and a closed-loop, continuous learning and improvement process.

How does ThinkLabs AI ensure the reliability and accuracy of its AI models in real-world scenarios?

Nature of physics-informed AI keeps AI grounded, tied to the real world, and bounded by the real world. We also do continuous learning and monitoring of model performance.

What makes your AI technology particularly suited for dealing with the complexities of modern electrical grids?

Being trained by determine engineering models, but handling the imperfect data quality of real world operations. AI also bring a wealth of optimization and generative techniques unmatched by traditional engineering mathematics.

How does ThinkLabs AI’s technology integrate with existing grid management systems like ADMS and DERMS?

ThinkLabs integrate as a Copilot with existing ADMS, DERMS, and AEMS, which will remain as the fundamental communications and control platform, while ThinkLabs will layer on additional intelligence and automation as similar to a vehicle's driving assistance system.