We don't pick sectors because they're hot. We pick them because we have conviction in how AI is reshaping them — and where the next decade of products will be built.
Consumer and prosumer tools where AI does the heavy lifting in the background. Voice interfaces that actually work, vision systems that handle real edge cases, productivity tools that don't just summarize your inbox but actually move things forward.
We're skeptical of AI products where the AI is the marketing layer on top of a regular product. The intelligence has to be load-bearing — the product should stop working if you remove the model.
The schools with the largest class sizes are also the ones with the most to gain from AI. We're building tutoring and content tools for classrooms that need to serve 25–40+ students at once — plus a dedicated line of work on early-childhood safety and engagement: how do you make screens for kids and toddlers meaningfully more educational, less addictive by design, and safer for young brains?
Our EdTech work is focused on North America and Europe — the markets where budgets, institutional buyers, and regulatory frameworks make AI-native tools an active buying decision rather than a philanthropy conversation. We also run a smaller Southeast Asia line where the same product primitives get stress-tested at higher class sizes and thinner infrastructure.
Physical devices with embedded intelligence. The thesis is that the next decade of AI products won't be chat windows — they'll be sensors, edge inference, and the supply chain to ship them at scale.
We work mostly with manufacturers in Shenzhen, Dongguan, and Suzhou. We prototype fast, run small pilot batches, and only commit to scale runs after we've seen real usage data. Most of our hardware bets spend 6-12 months in prototype before they see a single shipping box.
Wearables, sleep, mental health, longevity. The category is moving from step-counting toward continuous physiological monitoring and adaptive interventions. The winners will be the products that learn the user's patterns and surface what actually matters — not the ones with the most charts.
We're careful with health-adjacent products. Anything that makes clinical claims has to clear a regulatory pathway before it ships. We're not building diagnostic devices, but we build products that work alongside healthcare providers, not in place of them.
Consumer products for the home, kitchen, and everyday. We ship things that solve a small problem well, with AI handling the parts that benefit from intelligence (recognition, prediction, automation) and not pretending to be smart where it doesn't help.
Distribution for these products is selective DTC and a small set of retail partners. We're not trying to win Amazon. We're trying to build products that people buy once and tell two friends about.
Systematic strategies, market infrastructure, and the data pipelines that make quant work. The space is dominated by large funds, but there's a long tail of opportunities in niche strategies, regional markets, and infrastructure tooling that doesn't serve the Fortune 500.
We're careful about what we build here. Anything that touches market execution has to be measured against regulatory requirements from day one. We're not building black-box systems for retail — we're building the boring infrastructure that makes systematic strategies work at institutional standards.
The fastest way in is a short note. Tell us what you're working on, where in the journey you are, and what you need from a partner.