ABOUTThe Premise

A language for the outcome, not the data.

For thirty years, query languages have answered what happened. SmashGL was built to answer whether you're winning — by treating goals, windows, and visualizations as first-class objects, compiled together.

01Why now Four converging shifts

The conditions for a goal-first language did not exist until this decade.

Intent

Existing semantic layers define metrics — but they do not encode intent, thresholds, or success conditions.

Interpretation

Modern language models can disambiguate goal phrasing, recommend forecast models, and detect anomalies — capability that did not exist when SQL was designed.

Compilation

Analytics engines are now stable enough that a higher-level language can sit on top, compiling intent into existing query plans rather than replacing them.

Render

Visualization patterns — arc gauges, RAG alerts, sparklines with delta pills — have converged enough that they can be derived from a query rather than hand-built each time.

02What it is Four parts, one system

A query language is only the surface.

A · Language

Goal-aware syntax

Goals, windows, comparisons, and rendering live in the query — not in dashboard chrome built around it.

B · Compiler

Intent → query plan

Parses statements into an intermediate representation: metric plan, threshold plan, temporal aggregation, forecast model selection.

C · Engine

Goal evaluation

Computes progress, variance, status, and forecast. Operates in batch or streaming mode against one or more sources.

D · Renderer

Goal-aware visuals

Produces visualizations directly from intent — thresholds, progress, forecast curves, risk indicators — without manual dashboard authoring.

Existing analytics shows you what happened. SmashGL tells you whether what happened is what you wanted.
03Trajectory Where the language is, and where it's going

Stable surface today. Composable tomorrow.

v 1.0 · Today

Single-card grammar.

Ten clauses. Compiles atop GraphQL today; the design is engine-agnostic. Live SOURCE bindings or inline DATA / PREVIOUS for static cards.

v 1.1 · Near

Composite goals.

Hierarchical and multi-metric goals as first-class compositions. Goal graphs over multiple cards.

v 2.0 · Next

AI-assisted interpretation.

Ambiguous-goal disambiguation, forecast-model selection, anomaly detection — all part of the compile path, not bolted on after.

v 2.x · Beyond

Engine pluralism.

Adapters for additional analytics engines beyond GraphQL, including warehouse SQL dialects and streaming sources.

04Authorship

Designed in the open.
Patent pending.

The SmashGL language, compiler architecture, evaluation engine, and goal-aware visualization generator are the subject of a U.S. provisional patent application filed in April 2026.

The reference grammar is published openly so the syntax can be read, taught, and adopted by implementers. We welcome conversations about licensing, partnership, and reference implementations.