Nondeterministic Technologies is an applied AI company building products that do real work.

What NDT does.

NDT builds applied AI products. Each product is self-contained software that puts modern generative models to work on a real problem. The products on this site are real shipped or shipping software, not concept pieces. The portfolio grows as new bets get made.

Why it exists.

Modern generative models are good enough to build products that would have been impossible three years ago. NDT exists to take that capability seriously: to ship working products that put modern AI to work, learn from how they perform in actual use, and let those lessons compound into the next product.

The applied label matters. The company is not in the model-research business. NDT does not train foundation models, publish papers, or run benchmarks. NDT consumes the best models available and packages them into products people can use. The bet is that there is more value to be created at the application layer than is currently being captured.

How it operates.

NDT makes deliberate vendor choices. The technical stack is curated to minimize supply-chain risk and maximize control: Cloudflare for edge compute and hosting, Supabase for backend infrastructure, Anthropic for the deep generative work, with model-per-task selection (Haiku for short prompts, Sonnet for default generative work, Opus where reasoning depth pays for itself). The stack stays internally consistent so vendor relationships compound across products rather than fragmenting.

Who's behind it.

NDT was founded in 2026 by Meghneel Gore, an engineering leader with twenty-plus years in mobile software. More about the founder.