Meet Spotonix Analyst.
The AI analyst that interprets before it queries.

Ask a business question and inspect the analysis plan before SQL executes. Spotonix makes definitions, calculations, filters, and scope visible, then checks the resulting query against those material bindings before execution.

Spotonix is one product, not a family of loosely connected agents. Ask is the live core. Scheduled reports and alerts are live surfaces. Broader Insights and Reviews workflows carry explicit availability labels below.

The loop

One question, with the interpretation made visible.

The valuable part of an analytical answer is not only the chart. It is the business meaning underneath it: which population, calculation, time window, comparison, and filters the query used. Spotonix exposes that interpretation as a plan before execution.

Ask Live

Question → visible plan → answer

Ask in plain business language. Intent Algebra resolves the definitions, Calculations, scope, and filters into a visible plan. In Copilot mode, query generation pauses for your acceptance. The resulting SQL is checked against the accepted material bindings before it executes.

Scheduled reports & alerts Live

Put an agreed question on a schedule

Schedule reports and alerts for recurring analytical work. Broader watched-signal and automatically generated brief experiences belong to Insights and are not presented here as generally available.

Reviews Live

Decision and outcome memory

Reviews carries governed analysis into recurring meetings and preserves the decisions and outcomes alongside it, so each review starts from what the last one settled.

One product, one loop

1

Ask a business question

2

Read & approve the plan

3

Query is generated, checked & executed

4

Interpretation remains inspectable

Governed context informs later analysis

The Context Graph holds the business concepts and analytical relationships that ground interpretation. Where a workflow persists or reuses an accepted plan, that behavior must be demonstrated for that workflow rather than assumed from the product story.

Why it's different

It interprets before it queries.

A direct model prompt can draft SQL, but the prompt alone is not a durable representation of business intent. Intent Algebra binds the terms in a question to the Context Graph and exposes the intended analysis as a plan before SQL execution. In Copilot mode, you accept that plan before query generation continues.

Generate-first

The model turns a question into query text. Review happens at the SQL or answer layer, after the business interpretation has already been applied.

Interpret-first

The question becomes a plan you can read — Segments, Calculations, scope, and filters. In Copilot mode, you approve it before generation continues; validation checks material query bindings against the plan.

See the interpretation. Approve the plan. Inspect what runs.

Spotonix does not currently make a public guarantee that repeated runs produce identical SQL. The proof is the visible plan, the acceptance gate in Copilot mode, and validation of the material bindings used by the generated query.

See how a question becomes a plan → Intent Algebra

When it isn't sure

It asks instead of guessing.

Ambiguity is where analysis quietly goes wrong — which revenue, which date grain, which set of customers. When a question does not resolve to one plan, Spotonix surfaces the ambiguity and asks a specific question. In Copilot mode, the workflow pauses at the visible plan before query generation continues.

Illustrative — TPC-DS schema
You ask

"What were sales for our best customers last quarter?"

It asks back

"'Best customers' isn't defined yet. Do you mean the top 5% by lifetime spend, or repeat buyers in the last 12 months? And 'sales' — net of returns, or gross?"

Result

Your answer becomes a reusable Segment and Calculation — defined once, applied consistently, not re-guessed on the next question.

Deployment and controls

Know the boundary before you connect it.

Hosting, warehouse credentials, model-provider credentials, network path, and retained artifacts are confirmed for each deployment. Spotonix supports configurable model backends and warehouse adapters; it does not claim a universal runs-as-user identity model on this page.