Describe and design · Autessa Synthetic Data
Generate data from a description
Autessa Synthetic Data generates realistic data from a plain-English description. Write what the data is, what values are typical, and which edge cases you want, and Autessa turns it into typed, validated rows for a table or a dataset file.
The problem
Fake data usually means writing a generator
Demo environments, test suites and AI evaluations all need data that looks real. The usual options are hand-typed spreadsheets, field-by-field generators that produce obviously fake values, or scripts someone has to maintain. None of them capture the domain: the way support tickets are actually worded, which plans small customers pick, or the odd edge case that breaks your parser.
What you get
- Describe the domain, realistic value distributions and edge cases in your own words.
- The description drives every batch, so the tone and the mix stay consistent across thousands of rows.
- Column descriptions add per-column guidance, such as "mostly 30 or 45, sometimes 90".
- Optional sample rows act as few-shot examples for style and format.
- Generation is tuned for variety, and near-duplicates are removed across the whole job.
- The same description works for an existing table, a new table, a JSONL or CSV file, or a 50-row quick generate.
How it works
How does generate data from a description work?
Name the dataset
Give it a name such as "Demo customers" or "Support training set". For file targets the name is also the file name.
Describe the data
Write the domain, realistic distributions and the edge cases to include. The more specific the description, the better the rows.
Set the columns
Use an existing table’s columns, propose a schema from the description, or define the columns yourself.
Choose rows and write mode
Pick 1 to 5,000 rows and either "Review first" (recommended) or "Write directly".
Start generation
The summary line reads back exactly what will happen, for example "Generate 500 rows into inventory_items, staged for review." Then press Start generation.
Every option
What you can set when you describe a dataset
| Option | What it does |
|---|---|
| Name | Required. Shown in the jobs list; also the dataset file name for file targets. |
| Description | The plain-English description the job generates from. Required by the CLI and API, and needed to propose a schema. |
| Rows | 1 to 5,000 per job. Appends can add more, up to 5,000 across the job. |
| Sample rows | Optional JSON array of example rows used as few-shot examples. |
| Target | Existing table, new table, or dataset file (JSONL or CSV). |
| Write mode | Review first (stage for approval) or Write directly (write each batch as it is generated). |
Permissions
Who can do what?
| Who | Can |
|---|---|
| People in your company | Create datasets, see the company list of jobs, open any job and review, approve, append, cancel or discard it. |
| Table permissions | Rows written into a table are written as the person who started the job, with that person’s access to the table. |
| API keys | Can run quick generate, propose schemas and run dataset file jobs. Table targets need a signed-in person. |
Where AI does the work
How the generator handles this
The description is the prompt. Each batch is generated by your company default model with variety-focused settings, then checked against your columns, repaired once if needed, and de-duplicated before it is staged or written.
Generation runs on your company default model, and every row is checked against your columns before it is staged or written.
Your model and PII stripping →Connected
How generate data from a description connects to the rest of Autessa
FAQ
Questions about generate data from a description
How detailed should the description be?
As specific as you can make it. Name the domain, the typical values and their mix, and the edge cases you care about, such as "about one in ten invoices is overdue, a few have zero amounts". Vague descriptions produce generic rows.
Does the description leave Autessa?
It is sent to the model your company uses for synthetic data, which is Autessa built-in models unless an admin chose one of your own provider connections. It is stored encrypted with your company key.
Can I reuse a description for more rows later?
Yes. Append more rows to the job and Autessa reuses its description, schema and variety ledger, so the new rows match without repeating the old ones.
100% customizable, code and all
Make Autessa Synthetic Data fully your own.
Use it out of the box, then change anything. Ask Autobot, Autessa’s AI builder, for a new field, screen, report, workflow or agent and it builds it into your copy of the app. Or open the code and edit it yourself. Nothing is locked: the app, its data and its code are yours.
Start free and make it yours