Artificial intelligence is the connective tissue of nearly everything we build. We work on the models, agents, and learning systems that let machines perceive, reason, and act — and, just as much, on keeping that intelligence open, legible, and accountable to the people it serves rather than sealed inside a black box.
Wlcro brings consumer-grade fintech to the people and places incumbent banks treat as edge cases. We build payment rails, identity, and credit primitives that work first for the underserved — and turn out to work better for everyone.
Bio Net treats biological data as a network problem. We build the pipelines and shared infrastructure that let researchers model living systems collaboratively, with provenance and reproducibility built in from the first byte.
Mami is media built for the edge — content that renders, adapts, and is served close to the people consuming it. We work on low-latency delivery and machine-legible media metadata so both readers, human and machine, get a first-class copy.
TURF gives enterprises a defensible way to make hard decisions — structured, auditable models that turn scattered data and tacit judgment into choices a team can stand behind and revisit.
Auspicia brings the world's oldest spatial sciences — Vastu Shastra and Feng Shui — into a form a machine can reason over. We pair certified practitioners with spatial diagnostics that turn directional and elemental priors into clear, testable guidance for the spaces people live in and build. Every room has a direction; every direction has meaning, and we make that knowledge legible to both the homeowner and the algorithm.
Human body as a machine treats physiology as an engineered system — inputs, control loops, failure modes, and maintenance schedules. We model the body's subsystems with the rigour engineers bring to machines: where the levers are that keep it running well, and what the earliest signals look like when something drifts out of spec. The aim is a shared, falsifiable model of the body that a clinician, an engineer, and an agent can all read from the same page.
What does web infrastructure look like when the primary reader is a model, not a person? Indexing, provenance, and retrieval built for the second user.
Can the priors inside Vastu and Feng Shui be expressed as structured constraints a generative model can reason over — and tested against real outcomes?
Context-aware identity that travels across devices and the human/agent boundary. What does login mean when the "user" is an autonomous process acting for you?
Semantic diffing and bidirectional adapters so systems talk across versions without anyone rewriting the world every time an API moves.
How far can open, self-hosted LLMs go on real financial and planning tasks before cost or accuracy forces a frontier model? Benchmarks welcome.