Generate realistic data.
Export to any format.

SynthForge IO ships 111 field types across 19 categories: Basic primitives (integer, float, string, text, boolean, UUID, enum, date, datetime, integer sequences), Personal (first/last/full name, age, date of birth, gender, nationality, occupation, prefix, suffix, SSN, blood type), Contact (email, phone, area code, username, password), Address (street, city, state, postal code, country, building number, full address, secondary address), Geographic (latitude, longitude, license plate), Date & time (future date, past date, time), Business & company (catch phrase, company name, department, industry, sector, sub-industry, job title, vehicle), Commerce & products (order ID, product name, product category / subcategory / type), Financial (credit card, IBAN, SWIFT, currency code, currency amount, decimal, EAN, stock ticker, discount, price, percentage, tax rate), Internet & web (URL, domain, image URL, emoji, hashtag, slug), Networking & developer (IPv4/IPv6, MAC, port, HTTP method, MIME type, file extension, user agent, API key, locale, language code), Text & content (word, sentence, paragraph, book title, ISBN-13), Healthcare (CPT code/description, NDC code/drug/generic, ICD-10 code/description), Hardware (CPU, GPU), Food & drink (dish, drink), Education (academic degree, university), Color (color name, hex color), Measurement (height, weight), and Fantasy & creative (character class, item, name, race, spell, weapon).

SynthForge IO exports to CSV, SQL (DDL + INSERT statements for PostgreSQL, MySQL, SQLite, SQL Server, MariaDB, DuckDB, and CockroachDB), JSON, JSONL, and Apache Parquet. JSON / JSONL output is MongoDB-ready - load it via mongoimport. Each SQL export includes CREATE TABLE statements, bulk loading commands (\copy, LOAD DATA INFILE, bcp, etc.), and FK constraints optimized for the target database.

You can set value ranges, string length limits, regex patterns, enumerated values, distribution ratios, and date ranges on any field. Field dependencies use lookup tables to keep related values consistent (e.g., ZIP codes match their city and state). Derived rules let you define WHEN/THEN conditional logic with formulas that reference other fields.

Statistical distributions let you apply Normal, LogNormal, Exponential, Triangular, or Uniform probability distributions to numeric fields - so salary data clusters around a median, response times have a realistic long tail, or test scores follow a bell curve. Conditional weight overrides take this further: you can vary sampling weights based on another field's value (e.g., shift category weights by department, or product-tier weights by region).

Yes. For any child table that has a foreign key to a parent, the generation form lets you toggle from a fixed row count to a 'per parent row' ratio. Specify a min - max range (e.g. '8 to 12 patients per doctor') and SynthForge picks a uniform-random multiplier per parent. Ratios resolve in topological FK order, so a chain like doctors -> patients -> visits cascades automatically. Multi-parent ratios on the same junction table sum together (so '20 appointments per doctor' + '5 appointments per patient' produces a junction that respects both axes). Top-level tables stay as plain row counts.

Yes. Every successful dataset job persists its row counts and any ratios as the schema's defaults. The next time you open the generation form for that schema, the inputs pre-fill with what you last submitted. Override anything before regenerating; the new values overwrite the defaults. Works across browsers and devices since the defaults live on the schema record, not in local storage.