Added functionality to the beta Data Agent databases feature, including:
- Ability to generate Finish Personal Identity Codes (PICs)
- Ability to generate credit card data
- Ability to generate MongoID object identifiers
- Ability to upload .jpg and .png image files. For example, to provide a template for generating files from database data.
- Ability to generate PowerPoint files
- Ability to export the database to a JSON array
You can now use the Fabricate API to retrieve and download Data Agent databases.
From the Fabricate application, users can now delete their Fabricate account.
Users who try to sign up now receive a notification if the email address is already associated with an account.
The beta Data Agent feature now allows you to upload files as input.
The Admin Console now displays the Fabricate version number.
Fixed an issue where you could not delete a database if it had a variant.
Fixed an issue where new users on a self-hosted instance started a free trial.
Fixed an issue where users could not download exported data.
For the beta Data Agent database feature, added environment variables for self-hosted instances to provide an Anthropic API key and URL. The Data Agent uses Claude from Anthropic.
For the beta Data Agent feature, you can now prompt the Data Agent to use database data to populate PDFs, Word files, and email messages. For example, "Use the order data to produce example order confirmation email messages" or "Create an invoices database and create 5 example invoice PDFs". The Data Agent provides links to download the files.
Data Agent databases (Beta) - The Fabricate Data Agent is an AI chat interface that you use to create and define a set of data. You use the chat to make any adjustments to the data structure and content. The Data Agent asks questions as needed to clarify the request, then generates SQL and JavaScript to implement the changes. You can also run SQL queries to add or edit data, and can also edit values in individual cells. You can export the data from a Data Agent database to CSV, JSONL, SQLLite, SQL Server, or PostgreSQL.
When you create a database, Fabricate prompts you to decide whether to create a Data Agent database or a rules-based database, which is the original type of Fabricate database where you define rules for generating the data.
The Data Agent database feature is currently in beta and can be activated for specific Cloud customers. To request access to the Data Agent database feature, contact Tonic.ai.
The Professional plan is renamed to Plus. We also improved notifications and handling for free trials that are ending.
Fixed an issue with using the burndown option.
Fixed an issue where rows did not render correctly.
Fixed an issue where child fields could not refer to a field in a parent table because the field was not yet created.
The option to clone a database is moved from the database header to the Settings tab.
The Add Table(s) button is now a dropdown menu that contains the table creation options.
The new table creation menu includes a From SQL query option. This replaces the option to create what was once called a view, and is now referred to as a SQL table.
In the table creation menu, the new Generate static data using AI option allows you to create a static data table from an AI prompt. The new option generates a CSV file that you can download.
Self-hosted customers can now enable v2 signatures for Amazon S3. By default, Fabricate uses v4 signatures. To enable v2 signatures, set the environment variable FABRICATE_S3_SIGNATURE_VERSION=2.
Fixed an issue with setting the data type when exporting a column that contains numeric values, but that has a data type of string.
The new Markov Chain generator populates a column based on a configured set of available flows between values.
Fixed an issue with the display the Prevalence value for the Foreign Key generator.
Fixed an issue with the display of unstructured data values for CSV and JSON databases.
New Fabricate interpolation variables for SQL - You can now use the following variables when you use SQL in a column definition:
- {{parent(n)}}gets the nth parent of the column. Both {{parent}} and{{parent(0)}} retrieve the immediate parent.
- {{root_table}} gets the root table of an object or array.
For a Foreign Key generator that uses the Until column <= 0 distribution type, you can now configure whether to generate a foreign key for the first value of 0 or less.
Fixed an issue with importing tables when there were multiple spaces between CREATE and TABLE.
Self-hosted customers can now use Ollama as their LLM provider.
Streamlined the LLM provider environment variables:
- FABRICATE_LLM_URI replaces FABRICATE_OPENAI_URI and FABRICATE_AZURE_OPENAI_URI.
- FABRICATE_LLM_API_KEY replaces FABRICATE_OPENAI_API_KEY and FABRICATE_AZURE_OPENAI_API_KEY.
- FABRICATE_LLM_API_VERSION replaces FABRICATE_AZURE_OPENAI_API_VERSION.
The Credential Type setting for database connections is now correctly displayed only for Databricks.
When Fabricate imports a JSON schema, it now uses the minLength, maxLength, example, and description properties as input to create regular expressions to populate the columns.
Improved performance for deleting tables that contain columns with deeply nested data.
Fixed a JSON parsing issue.
A new Person Name Title generator applies either Mr., Mrs., Ms., or Dr. as the column value. Dr. is applied to 1% of the records, with the other values distributed evenly and corresponding to the gender of the name columns.
A new Person Name Suffix applies either Jr., Sr., or III as the column value. The suffix is applied to 2% of male names.
When you export to a Databricks database, you can now choose whether to authenticate using a service principal or a personal access token.
For databases other than XML and JSON, the database details now include a DDL tab that contains the DDL used to create the database. You can update the DDL, and choose whether Fabricate updates the database to reflect the revised DDL. When Fabricate updates the database, it does not touch hidden tables and columns. When you export the database, you can choose whether to include the DDL.
For a PostgreSQL or SQL Server database, you can now optionally provide a schema name. The schema name field is on the database Settings tab. When you export the database to SQL or to a database, the database tables are all assigned to that schema.
Performance improvements.
Added a user-agent header to Databricks API calls.
When you create a database or add a table, a new AI Hints field allows you to provide an additional prompt to help further define the database or table. For example, you might indicate specific generators to use or specify limits on column values.
Fixed an issue with importing fields containing an update date.
Fixed an issue with array value uniqueness.
Added the following generators to produce values related to a person's job - Job Area, Job Descriptor, Job Title, Job Type.
Added the following generators to produce passport numbers - Passport Number (Global), Passport Number (US).
Fixed an issue with exporting to a Databricks database.
Fixed an issue with generating longitude and latitude values.
Fixed an issue with creating a database from an XSD file.
You can now create and export to a Databricks database.
You can now export a database to XML. Fabricate creates a .zip file that contains an XML file for each table.
The new Array generator populates a column with an array of values of the same type. You select the generator to use to create the values, and configure a distribution of array lengths.
The new Object generator populates a column with a JSON object. You configure the list of fields. For each field, you select the generator to use to create the field value. You can optionally import a JSON schema instead of configuring the fields manually.
You can use the Object and Array generators to produce a nested JSON structure.
When you create a new database, you can now provide a prompt that describes the database, or a definition file such as JSON or XML. Fabricate's AI can then use that to generate the database schema.
The new Middle Name generator generates a name value, similar to a first name, that matches the gender of the original name.
Fixed an issue where error messages for views did not display.
Fixed an issue where the database configuration to set the row count based on unique values was not saved.
Invited users can now use SSO to accept the invitation and create their account.
Fixed an issue where the group key field was missing for Credit Card Number and Credit Card Type.
You can now configure a Number generator variant to be based on a percentage condition instead of a SQL expression. For example, you can configure a variant to be used in 5% of the rows. A new condition type option allows you to select whether to provide a SQL expression or a percentage as the variant condition.
For the Lorem Ipsum generator, the distribution options are now limited to Fixed, Uniform, and Normal.
Security enhancements.
For the Unstructured Data generator, we no longer display the Output Format field, which didn't apply.
The new Sum From Another Table generator populates a column based on the sum of values for a column in a set of rows in another table. For example, in an orders table, the order_amount column is populated with the sum of the price column in the order_items table, for rows where order_ items.id matches orders.order_id.
To avoid confusion with the new table, changed the name of the Sum generator to Sum Of Previous Rows.
When you purchase a Fabricate license, you must now provide a billing address.
Fixed an issue where an incorrect quote character was used in exports to MySQL.
Security enhancement.
Changed the configuration options for Binomial distribution. You are now prompted for Trials and Success Probability. Removed Minimum, Maximum, and Decimals.
Improved how Fabricate derives List values.
Enhancements for improved security.
Fixed an issue where the Uniform distribution never used the specified maximum value.
On the cloud application, you can now select the LLM provider to use for the Unstructured Data generator. You can also set the model to use for each type of content (free text, conversation, and JSON). The configuration is available to account owners, and is in a new Unstructured Data Generator section of the My Account page.
For additional security, Fabricate now limits login attempts from a single IP address to 5 per minute, and locks out users after 5 failed login attempts. Self-hosted instances can configure these limits using the environment variablesFABRICATE_MAX_LOGIN_ATTEMPTS_PER_MINUTE and FABRICATE_MAX_LOGIN_ATTEMPTS.
Improved performance for data regeneration.
Creating a database based on Swagger YAML - On the database creation panel, in the Schema Definition field, in addition to a SQL DDL script or a prisma.schema script, you can now provide Swagger YAML API definition. Fabricate uses the API endpoint request and response information to derive the database tables and columns.
For the Foreign Key generator, the cardinality distribution types are now limited to Fixed, From Column, Normal, Uniform, and Until column <= 0.
Self-hosted instances can now use either Azure OpenAI or OpenAI to support Fabricate AI features.
Made general improvements to user interface text.
Added the following functions to allow you to generate Luhn check digits:fab_luhn_check_digitfab_append_luhn_check_digit
Added the ability to generate HCPCS codes.
Added the ability to generate US Hospital data.
Added the ability to generate US Social Security Numbers.
Added the ability to generate US Employer Identification Numbers.
Added the ability to generate US Bank routing numbers, names, and addresses.
Added the following new function to add typos to a string:fab_add_typos(input, unit, value)
When you use the Number generator, you can now specify conditional variants. For example, you could use this feature to create seasonal pricing trends.
Fixed an issue where Percentage Null was ignored for Foreign Key fields.