On AI, power electronics, venture building, and the industrial future.
The most common mistake in industrial AI adoption is treating intelligence as an afterthought. Companies retrofit ML models onto unchanged hardware architectures and wonder why the performance gains are marginal. Here's why that approach fails at scale — and what the alternative looks like.
Read More →Conventional drives are reaching the limits of analog control theory. Variable-frequency drives today are more capable than ever — but their fundamental architecture hasn't changed in decades. AI-native drives represent a step-change in what's possible for torque control, efficiency optimization, and fault response.
Read More →Hardware has a reputation for being capital-heavy and slow. That's often true — but it's also what makes the moat defensible. Software-only AI companies can be replicated in months. An AI-native industrial drive takes years of domain knowledge to get right. Here's how I think about the tradeoff.
Read More →The decision to start a deep-tech hardware company is not one you make lightly. It requires conviction about the technology, clarity about the market, and realistic expectations about the timeline. This is the story of how Sparrow Quest came to be — and the problem it's built to solve.
Read More →Cloud-connected AI makes sense for many applications. For industrial control systems, it often doesn't. Real-time control demands microsecond response times. Safety-critical systems require deterministic behavior. Air-gapped facilities need operational continuity. Edge AI is not a compromise — it's the right architecture for the domain.
Read More →