- Essays··8 min read
The Compute Bill Sitting Under J-MID
Japan and Korea are deploying AI across every teaching-hospital radiology workflow — yet neither ministry nor hospital consortium has published the electricity budget for the inference stack. Continuous report drafting, MRI classification and admission triage draw power differently from the scanners they augment, and the gap between an H100 rack and a Furiosa chip is a hospital-substation decision. The operating energy assumption under each country's AI hospital programme remains unwritten.
radiology-aimedical-imagingcompute-costshealthcare-aiRead - Essays··8 min read
The Underwriter Reads Second
The safest AI architecture in insurance underwriting does not let the model decide — it lets the model read. Across every serious deployment in Japan and Korea in 2026, the machine extracts, classifies, and routes; a named underwriter signs the outcome. Japan's FSA and Korea's FSC arrived at this split from different premises — explainability on one side, fraud containment on the other — and reached the same architectural conclusion.
insurance-aiunderwritinghuman-in-the-loopfinancial-regulationRead - Essays··8 min read
Two Ledgers in a Kagome Field
Kagome's AI yield model forecasts harvests five weeks out from soil sensors and satellite pixels; its SSBJ-mandated Scope 3 ledger must account for the same fields in itemised grams of CO2-equivalent. The two models share no instrumentation — a nitrogen prescription that reduces N2O flux does not automatically become a credited emissions delta. That reconciliation is where Japanese food processing will spend the next several years, regardless of whether precision agriculture's sustainability claims hold up.
precision-agricultureclimate-disclosurescope-3-emissionsjapanRead - Essays··8 min read
The Rice Paddy Needs a Planner: What Japan's Hillside Robots Say About Agent Design
Japan's hillside farms cover 40% of cultivated land, with fragmentary cell coverage and a workforce averaging 69. They force a reckoning with agent architecture that enterprise AI teams have deferred. When the link is intermittent, the planner-executor split is the only arrangement that lets the tractor stop safely. A physically independent critic completes the trio, and no board should approve an architecture that collapses all three.
agent-architectureagricultural-roboticsplanner-executorfunctional-safetyRead