The who
Segments
The cohorts, store sets, and product groups your team refers to by name — high spenders, flagship stores, lapsed accounts. Defined once, so the name can resolve to an explicit governed definition instead of an unstated assumption.
Product · Context Graph
The meaning behind the analysis,
explicit — and inspectable.
Spotonix is the AI analyst that interprets before it queries. The Context Graph is where that interpretation is kept: the definitions your team governs — Segments, Calculations, and Analysis Patterns — together with schema relationships and other context used to interpret analytical questions.
It makes business meaning explicit and inspectable instead of leaving it scattered across prompts, query fragments, dashboards, and individual memory.
What the graph holds
Not a data copy — the accepted meaning.
The Context Graph does not store your rows. It stores the agreements your team makes about what the numbers mean. Four kinds of context can be represented in it.
The how much
Calculations
Growth, ratio, rate, benchmark, and time comparison — the arithmetic your team has agreed is correct. When someone asks for year-over-year growth, the intended definition can be shown in the analysis plan.
The reusable intent
Analysis Patterns
The semantic glue that binds Segments and Calculations into a repeatable analytical shape — a way to represent the analytical operation separately from the wording of a single question.
The analytical trace
Plans & Answers
Plans and Answers provide inspectable context for an analytical interaction. Persistence and reuse depend on the active workflow and should be evaluated directly rather than inferred as a universal behavior.
How the graph is built
It starts from what you already have.
The graph can be seeded from warehouse schema and supported metadata sources, then reviewed and refined for the deployment. Connector coverage is documented on the Integrations page.
Schema
Tables, columns, keys, and joins — the shape of the warehouse
Query context
Examples and relationships supplied for the deployment
Metrics
The measures your business already tracks and trusts
Supported assets
→
Definitions from connectors confirmed for the deployment
Reviewed context
Explicit meaning available to interpretation
These sources ground Intent Algebra. The Context Graph supplies explicit domain context; Intent Algebra composes the visible analysis plan from it.
Why it matters
Business meaning should outlive a prompt.
A prompt is transient. A governed definition should be explicit, reviewable, and available to the next analytical interaction. The Context Graph separates that durable domain context from the wording of any one question.
Without explicit context
Meaning stays scattered
- Definitions live in dashboards, SQL, and team memory
- A prompt has to reconstruct the intended meaning
- Review begins only after a query or chart appears
With the Context Graph
Meaning is available to inspect
- A Segment resolves to an explicit definition
- A Calculation can be shown before execution
- An Analysis Pattern helps compose the plan
Illustrative walkthrough on TPC-DS data — not a benchmark. Actual reuse depends on your graph.
The claim is inspectable meaning, not identical SQL. Persistence and reuse are demonstrated per workflow; Spotonix does not present them as an automatic guarantee on this page.
The mechanism that turns a question into that reusable plan is Intent Algebra. The Context Graph is what it composes from; Intent Algebra is what does the composing.
Where the boundary is
Governed meaning — not a second copy of your data.
The most common misread of a "graph" is that it is another place your data lives. It is not. Being precise about what the Context Graph is keeps the evaluation honest.
- It is not a replacement warehouse. It models analytical meaning and relationships. The exact artifacts retained and where they are stored are documented for the selected deployment.
- It is not model weights. Your business meaning is not baked into a fine-tuned model; it is explicit, inspectable structure in the graph — the same meaning your team can read and change.
- It is not a free-form prompt history. Meaning is represented as explicit concepts and relationships that can be inspected and corrected.
- It is not Intent Algebra. The graph is the durable memory; Intent Algebra is the mechanism that interprets a question and composes a plan from that memory. One remembers, the other reasons.
Controls & ownership
Your meaning, with its boundary documented.
Deployment location
EngagementStorage location, network boundary, backup behavior, and retained artifacts are confirmed for the selected deployment. Do not infer a universal topology from the product model.
Inspectable, not opaque
LiveSegments, Calculations, and Analysis Patterns are explicit definitions your team can read, review, and correct. When meaning is wrong, you change the definition — not retrain a model.
Accepted, not assumed
LiveIn Copilot mode, a person can review and accept the visible plan before query generation continues. Other persistence and approval flows are documented by workflow.
Graph portability
LiveThe graph is your asset. Open export in a standard, non-proprietary format so it leaves with you is in development — we will confirm the format and scope with design partners rather than overstate it here.
How the graph is stored and governed is confirmed per engagement, against your real environment — we work this through with design partners, not a reference diagram.
Keep reading