Boltzbit AI

Dr. Yuchuan Zhang's passion for AI began during his PhD at the University of Edinburgh in 2010, where he published research on Boltzmann machines and approximate inference at top AI conferences. This deep technical foundation would become the backbone of Boltzbit AI. The core insight was simple but powerful: most businesses have valuable data but lack the machine learning expertise to extract value from it. Zhang decided to build a platform that would democratize AI, letting anyone with data build intelligent systems without needing to be a machine learning expert.

Zhang started writing code in April 2020, initially working without salary alongside two co-founder friends for roughly 4-5 months to build prototypes. He held the majority equity stake due to leading the initial vision. The early traction earned them a pre-seed round of $800K in 2020, followed by a $1.6M seed round (likely in 2021) at a valuation between $10-15 million—roughly 15-20% dilution, which Zhang felt was fair for a deep-tech company at such an early stage.

The first customer came through personal connections in a specific niche vertical where the team deeply understood the problem. They landed a $130K annual contract (approximately $12,500 MRR) for a search engine use case that leverages generic TVI to parse documents combining text and images, allowing users to interactively search and highlight relevant sections. This was a significant validation of their product-market fit for at least one vertical.

Zhang credits their success to finding a problem they truly understood and solving it for customers in that space. Rather than spreading thin, they validated deeply in one vertical before expanding. Their business model—usage-based pricing tied to data uploads and computational hours—aligns customer success with company growth. Now with 9 full-time employees (3 co-founders plus developers and business staff), they're actively exploring adjacent verticals in fintech and digital marketing through a combination of cold outreach and partnership discussions.

Boltzbit has one paying customer generating $150K ARR and is not currently fundraising, having recently closed their seed round. They're exploring multiple verticals while remaining selective about hiring, planning to scale the team as new customer wins materialize. The platform's core technology—built on public datasets like COCO and ImageNet—proves that you don't need proprietary data to build powerful AI solutions. Their next chapter will reveal whether their search technology can dominate adjacent markets or if they'll need to adapt the platform for different use cases.