The latest articles on DEV Community by yocho — Hikaru Sato (@hikaru_sato_ce8ce2cff7c01). https://dev.to/hikaru_sato_ce8ce2cff7c01 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4072584%2F5b310885-b077-4a80-8c6f-73de113d6dec.jpg https://dev.to/hikaru_sato_ce8ce2cff7c01 en yocho — Hikaru Sato Wed, 12 Aug 2026 03:42:03 +0000 https://dev.to/hikaru_sato_ce8ce2cff7c01/why-im-building-yocho-from-events-to-hypotheses-2h9e https://dev.to/hikaru_sato_ce8ce2cff7c01/why-im-building-yocho-from-events-to-hypotheses-2h9e <p>AI industry news is abundant, but information alone does not make the next change easier to understand.</p> <p>I’m building yocho.ai, an AI industry intelligence system that keeps four layers distinct:</p> <ul> <li>Event: what happened</li> <li>Entity: who or what is involved</li> <li>Relationship: how changes connect</li> <li>Hypothesis: what may follow and why</li> </ul> <p>This boundary is practical. Facts can be corrected, relationships can be re-evaluated, and hypotheses can change as new evidence arrives. An AI system should help organize the evidence without silently turning its own interpretation into an official conclusion.</p> <p>I’m starting with a founder build log covering design decisions, implementation lessons, discarded approaches, and open questions.</p> <p>If you are interested in evidence-led AI industry intelligence, I’ll share more as the system develops.</p> <p>Learn more: <a href="https://yocho.ai" rel="noopener noreferrer">https://yocho.ai</a></p> <h2> A concrete example </h2> <p>For illustration, imagine Company A releases Model B. yocho keeps the layers separate:</p> <ul> <li> <strong>Event:</strong> Company A releases Model B.</li> <li> <strong>Entity:</strong> Company A, Model B, and competitor Company C.</li> <li> <strong>Relationship:</strong> Model B may change Company C’s pricing strategy.</li> <li> <strong>Hypothesis:</strong> Lower inference costs may change how teams evaluate adoption.</li> </ul> <p>A simple flow is: <strong>Source → Event → Entity / Relationship → Hypothesis → Review</strong>. The hypothesis is provisional, not a fact. New evidence can support it, weaken it, or leave it unresolved.</p> <h2> What’s next </h2> <p>I’m using <a href="https://yocho.ai" rel="noopener noreferrer">yocho.ai</a> as the canonical hub for the product’s design decisions, evidence, and implementation updates. The next useful step is to compare a hypothesis with the sources and relationships that support it.</p> opensource