How the AI analyst interprets before it queries. This paper lays out Intent Algebra: the mechanism that resolves the definitions, calculations, scope, and ambiguity in a business question into a visible plan — and only compiles SQL once that plan closes.
A walk through the reasoning behind Spotonix — why generating SQL directly from a question is fragile, and what it takes to interpret intent first. Each section is written for technical and analytics leaders evaluating how the system actually works.
Why text-to-SQL leaves a semantic gap: business logic buried in joins and filters, no reuse across related questions, and reasoning you cannot inspect. The case for interpreting intent before touching the warehouse.
Analysts do not start from scratch. They recognize what a question resembles, then reuse and adapt prior work. The paper formalizes that compositional way of working.
Any question maps to a few durable pieces — the who (Segments), the how much (Calculations), and the reusable analytical intent that binds them (Analysis Patterns).
Where definitions, accepted plans, and how the business uses them are remembered and connected — so meaning compounds instead of being re-derived for every question.
The mechanism that binds the terms in a question, grounds them in the Context Graph, surfaces ambiguity instead of guessing, and gates compilation on a closed, visible plan.
How the system searches, evaluates, asks clarifying questions where the intent is underspecified, and — once a plan is accepted — turns it into an Answer that can be reused.
The whitepaper is shared with teams evaluating Spotonix. Talk to the founders and we will walk you through the approach on your own data.
