Getting started

The budget#

Section 13 of the PRD sets a design budget: under 3 minutes from finishing installation to a generated file. It is stated as a budget, not a measured median — measuring a median would require instrumenting users, which Section 10 forbids — and it is verified by hand, on the maintainer's own machine, with no account, no key, and no network.

This page is that walkthrough, using the app's own real, already-built fastest path. The PRD also names schema import from an existing CSV/Parquet/SQL-DDL/JSON-Schema file as "the strongest available lever on the time-to-first-dataset budget"; that lever is not built yet. Until it is, Scenario Mode's template tier is the fastest real on-ramp, and is what this page uses.

1. Launch#

The Builder opens with one table (table1, 100 rows) and one column already typed (id: identifier.sequential) — never a blank screen.

2. Describe what you want#

Click Scenario and type a plain-language description of the columns you want. This recipe used "student id, grade level, gpa, enrollment date":

Answered by: local above the match list names which of Scenario Mode's three tiers actually answered this draft (see Scenario Mode for what the other two look like). Click Apply and confirm to replace the working recipe with the draft.

3. Export#

Click Export, confirm the destination folder (it defaults to the Settings-configured project library), pick a format, and click Export:

That is a generated file on disk — CSV by default, or any of the other seven §5.6 formats. Every export also writes a data dictionary and a provenance record beside the data unless you turn them off in this same dialog.

Finding things faster#

⌘K opens the command palette from anywhere in the app — a single search over top-level commands, every catalog type, and every Settings control:

Where to next#

• Recipe files — save this recipe to reopen later, or hand it to sensym generate from the CLI.
• Recipe library — a live view over your saved recipes: open, duplicate, reveal, or delete any of them.
• Scenario Mode — all three tiers, and what each one needs to run.
• Cohort generation — generate one seeded dataset per row of a class roster in a single run.
• The causal graph view — declare a treatment/outcome/confounder structure and read its true, closed-form ATE.
• Speech narration — install the real neural TTS runtime and read a text column aloud into real WAV files.
• Type catalog reference — every registered type, grouped the way Appendix C groups them, with its own real example and parameters.
• CLI reference — the same sensym command line documented start to finish.