Human first
AI is the engine, not the product. Every feature exists to give a person a capability they did not have yesterday.
Assistive · applied AI
We build products that put advanced perception, prediction, and generation to work for people in the field: beekeepers, event-goers, neighbours watching the sky, and everyone sharing the road. One shared platform underneath, real verticals on top.
AI is the engine, not the product. Every feature exists to give a person a capability they did not have yesterday.
Identity, following, feeds, notifications, media, and weather are shared. Each product is a thin vertical over that core.
Phones, sensors, cameras, and weather. Our work runs where the data is: on device, at the edge, and on GPUs when it earns it.
Projects
Products for specific communities, from beekeepers to people on the road, plus the tooling we use to build them.
Hive and apiary management with AI photo analysis for colony health, pests, queen presence, and brood pattern. Commercial beekeepers broadcast their inspections, feedings, and treatments so nearby smaller operations can see what to do and when.
A hyper-local events platform with AI recommendations. Venues, ticketing integrations, multimodal embeddings for discovery, generated flyers, and payments, so people find what's happening near them tonight.
Protecting vulnerable road users. Phone sensors and camera detect hazards on the move; the backend fuses text, image, sensor, and geometry embeddings, reconstructs scenes in 3D on GPU, and maps surface hazards tile by tile.
One place for a family to understand and manage a child's complex medical situation: a unified timeline of visits, labs and diagnoses, tracking across conditions rather than silos, a full history of what has been tried, and controlled sharing with the specialists who need it. Local-first and end-to-end encrypted, so records stay on the family's devices and nothing leaves unless they share it.
Remember everything, build better. One connected record of the software lifecycle: product requirements, UI design, technical design, and QA, all linked and versioned. AI reviewers check security, privacy, compliance, and performance, and change-impact tracking shows the ripple effects of every change before it ships.
Platform
One company, one infrastructure. Every product imports the same Python package and registers its own records as feed sources. The record is the content: an inspection, an event, a weather report, a hazard. No copied social code, no per-app ops stack.
Build on it with usConsulting
We take on a small number of engagements where applied AI has to work outside the lab.
Vision models on device, multimodal embeddings, retrieval, and LLM-backed features that hold up in production.
Kotlin Multiplatform apps for Android and iOS, high-rate IMU and camera pipelines, positioning, and geospatial data.
Shared substrates for product families: identity, feeds, eventing, GPU dispatch, and the ops setup to run them lean.
Contact
Consulting, partnerships, or a question about one of the projects.