Glossary

A quick reference. Each entry links to the page that covers it in full.

Getting your data in

Source — A connection to a system you already use: Shopify, HubSpot, Stripe, an ad platform, a spreadsheet. Kept in sync automatically. → Sources

Sync — The scheduled copy of data from a source into your warehouse. Each source reports when it last ran and whether any table failed.

Warehouse — The single place all your data lives. Included and managed; you never size or tune it. → Storage

Raw — Data exactly as the source delivered it. Never edited by hand. The audit trail behind every number.

Shaping it

Model — Your business's definition of something: an order, a customer, a campaign. Built from raw data, and the thing every chart and answer reads. → Models & Semantic Layer

Semantic layer — The models plus the meaning attached to them: what each property represents, how each metric is calculated, and how models relate. What documents, Operator, and automations all query.

Property — A field on a model, one per column the model's SQL selects. Typed and described, so the meaning travels with the column.

Aggregation — A metric defined once on a model — sum_net_revenue, count_orders — so every chart agrees on what it means.

Relationship — A defined link between two models, so you can combine them without writing a join.

Build run — One pass of building your models, in dependency order. Shows what ran, what changed, and which model failed.

Graph — The dependency graph of everything in your workspace: sources, models, and every document and automation that reads them. Useful for seeing what a change affects. → Observability & Lineage

Using it

Document — A page where text and live data sit together. What other tools call a dashboard. → Documents & Explores

Block — One element in a document: text, a chart, a table, a filter.

Explore — The block that queries a model. Pick properties, aggregations, and filters; show the result as a table, chart, scorecard, or pivot.

Scorecard — A single metric shown large, usually with a comparison to the prior period.

Pivot — A cross-tab view, for two dimensions at once.

Operator — Switchboard's AI. Answers questions, runs analysis, builds models and dashboards, and sends what you ask it to. → Operator & Model Harness

Model harness — The structure around the language model: the context it is given, the tools it can run, and the approval it needs before acting outside Switchboard.

Automation — Work that runs on a schedule or a condition and delivers a result somewhere. → Automations

MCP — Model Context Protocol, the standard Switchboard uses to expose your models to an outside AI client. → Connect Claude (MCP)

Around the edges

Home — Your landing page: pinned documents, plus an onboarding checklist if you're new.

Settings — Your organisation's configuration: name, business timezone, start of week, and billing.