Analytics & Search

Self-serve analytics has been promised for two decades. What broke it every time was the gap between question and execution - the unspoken assumptions buried in column names, the join paths that only the data engineer remembered, the metric definitions that drifted across dashboards. AI didn't fix this gap. AI exposed it.

This collection examines what changes when analytics and search are both grounded in a typed semantic graph. Posts cover multi-hop query understanding: how an agent navigates from "Q3 NPS for the lapsed-renewal cohort" through entities, relationships, and proven join paths to a single deterministic SQL plan. They also cover semantic search across corporate data - moving past keyword and embedding-only retrieval into context-aware concept resolution, where the same word resolves differently for finance, sales, or legal.

You'll find pieces on conversational analytics for business teams that don't speak SQL, and on how the same semantic substrate that powers search also powers governed dashboards, agent workflows, and audit. The thesis: when analytics and search compile through the same graph, the answer is not just relevant - it is provably correct, policy-aware, and traceable to the business logic that produced it.