§01|SaaS

--profile saas


Bias generated values toward a B2B SaaS product: workplace emails, startup-style company names, plan-tier enums, and dates clustered in the last 18 months.

prompt hints
  • startup and company names mixing single-word brands with "X-ly" and "Get-X" patterns
  • workplace email addresses (alice@acme.com), not gmail / hotmail / personal domains
  • plausible SaaS job titles (Engineer, PM, Designer, Head of Ops)
  • billing-plan and feature-flag enums when the column name suggests them (free, pro, team)
  • dates clustered in the last 18 months; subscriptions distributed across plans
sample values
columnexample
orgs.nameNorthbeam Analytics
users.emailalice@northbeam.io
users.titleStaff Engineer
subscriptions.planpro
users.created_at2025-11-04T09:22:11Z

Profiles bias value choice; they don’t define your schema. The table set, column types, and FK graph come from introspecting your own database. Rows are validated against your Postgres constraints before they land.

§02|E-commerce

--profile ecommerce


Bias generated values toward a consumer e-commerce store: realistic product names across apparel/home/accessories, prices with familiar retail cents, and orders skewed toward fulfilled.

prompt hints
  • product names across apparel, home goods, and accessories
  • prices between 9.99 and 299.99 with cents ending in .99 / .49 / .00
  • customer names and shipping addresses drawn from US / UK / DE / FR / JP
  • order statuses biased toward fulfilled, with a long tail of pending and refunded
sample values
columnexample
products.titleCedar Plank Cutting Board, 18"
products.price$34.99
customers.countryDE
orders.statusfulfilled
orders.total$84.50

Profiles bias value choice; they don’t define your schema. The table set, column types, and FK graph come from introspecting your own database. Rows are validated against your Postgres constraints before they land.

§03|B2B

--profile b2b


Bias generated values toward a B2B service or marketplace: mid-market accounts, contract values, multi-seat licensing, and procurement-style metadata. Consumer language is avoided.

prompt hints
  • mid-market company names with a named-account feel
  • contract values in the $5,000 to $250,000 range
  • multi-seat licensing (seat counts, per-seat unit prices)
  • procurement metadata: PO numbers, NET-30 terms, MSA dates
  • no consumer language (no shopping, no household products)
sample values
columnexample
accounts.nameMeridian Freight Systems, Inc.
contracts.value_usd$78,000
contracts.seats45
contracts.payment_termsNET-30
contracts.po_numberPO-2026-01849

Profiles bias value choice; they don’t define your schema. The table set, column types, and FK graph come from introspecting your own database. Rows are validated against your Postgres constraints before they land.

§04|Bring your own

no profile? no problem.


Omit --profile and the CLI runs with a neutral system prompt. Values still respect column types, nullability, unique constraints, check constraints, and foreign keys—you just don’t get the domain-flavored biases. Useful for internal schemas that don’t map cleanly onto SaaS, e-commerce, or B2B.

More profiles (legal, real estate, logistics, healthcare) are on the roadmap and will be pinned by user demand. Open an issue with your CREATE TABLE statements and a short description of the domain and we’ll triage.