Test data that looks real. Because it isn't.
DataFactory gives every test environment production-shaped data, without a single real customer in it. Cut a huge database down to size with every relationship intact, or generate realistic data from nothing but a schema.
Self-hosted on your own server with Docker. Nothing is sent to Enov8.
Prune output is real data, just less of it. Pair it with masking where needed.
| MEMBER | MEDICARE | IHI | TFN |
|---|
Values above are generated in your browser with real check-digit rules. Example plan only.
01Two engines
Shrink what you have. Grow what you don't.
Prune cuts real databases down to a size your test environments can handle. Bloom builds realistic data from scratch. Use either, or both.
Subsetting
Cut a large database down to anywhere from 1% to 100%, with every foreign-key relationship intact. Nothing orphaned, nothing broken.
- Any target size, plus optional row filters by value, text, year, number range, or empty / not empty
- Read-only on the source. Prune never changes it
- Reduction script that shrinks a cloned copy in place
Synthetic data
Generate realistic synthetic data from a schema. No source rows needed, so no real customer is ever involved.
- 250+ skills with real check digits, from card numbers to Medicare and NHS numbers
- Relationships set by rows per parent, with uniform, Poisson or long-tail distributions
- Business rules applied while generating and checked afterwards
- "Learn from source" copies only table sizes and value mixes, never personal values, and drops rare values
- Up to 10 million rows per run by default. Large runs work on disk, not in memory
02How it works
From source to test environment in four steps.
Define a data set once and reuse it for every prune and every plan.
Connect
Point DataFactory at a source database with a read-only account. Or skip that and start from a SQL schema or a template.
Define
Pick your anchor tables. DataFactory maps every related table into a reusable virtual data set.
Prune or Bloom
Cut a subset to size, or generate synthetic rows with the skills, volumes and rules you choose.
Deliver
Download files, load straight into a test database, or let your pipelines and AI agents do it.
03250+ skills
Fake data that passes real validation.
Skills are value recipes. Identifiers come out with real check digits, so they get past the same validation your applications run in production.
Try it: every value on the right is generated in your browser using the same public check-digit rules.
Synthetic values for testing only. They match the format rules, not real people.
04Industry packs
Twelve templates that speak your industry.
Start from a template built around the identifiers, records and events your systems actually use. Admins switch packs and regions on or off.
Banking & finance
1 templateAccounts, cards and payments with valid bank identifiers.
Health insurance
4 templatesAU, UK and US member and provider data, including a template aligned to the Ensure health-fund core system.
General insurance
1 templatePolicies, vehicles and business customers.
Telecommunications
1 templateSubscribers, devices, SIMs, usage records and number porting.
Government & public sector
1 templateCitizens, payments, cases and grants.
Energy & utilities
1 templateMeters, tariffs, reads, bills and concessions.
Superannuation & wealth
1 templateMembers, contributions, rollovers and insurance in super.
Retail & e-commerce
1 templateCustomers, products, orders and fulfilment.
Salesforce
Platform · 1Sales Cloud objects with properly formed record IDs.
Admins choose which packs and regions are switched on for their organisation.
05Quality you can prove
Realistic shape. Enforced rules. Checked every run.
Bloom doesn't just fill columns. It builds relationships with realistic spread, applies your business rules as it generates, then validates the result.
A quality score on every run
Example run shown.
Relationships with realistic spread
Set rows per parent and how they're distributed.
Business rules that hold
Applied while generating, then checked afterwards.
same plan + same seed= the same data, every time. Reproduce a failing test months later, row for row.06AI & automation
AI if you want it. Never your data.
DataFactory works fully without AI. Switch it on and you can describe columns, rules and changes in plain English. Only the structure goes to the model.
✓Sent to your AI provider
- Table and column names
- Plan settings
- The project brief
✕Never sent
- Rows of data
- Source values
- Database passwords
Built for pipelines
A REST API covers prunes, Bloom plans, runs and loads, so fresh test data becomes a step in CI/CD.
Ready for AI agents
The built-in MCP server lets agents such as Claude work as a signed-in user. No database passwords pass through the agent.
07Security
Runs inside your walls.
DataFactory is self-hosted on Docker. Your data never leaves your network, and nothing is sent to Enov8.
08Better together
Part of the Enov8 platform.
DataFactory slots in next to the Enov8 tools you may already run.
Mask what you keep. Replace what you don't.
Test Data Manager masks the real copies you need to keep. DataFactory replaces the ones you don't.
Explore Test Data Manager →DataFactory + VirtualizeMeClone in seconds, then shrink.
Clone a database in seconds with vME, then run DataFactory's reduction script to shrink the clone.
Explore VirtualizeMe →DataFactory + Env & Release ManagerBook it. Then fill it.
Book the environment in Environment & Release Manager, then fill it with exactly the right data.
Explore Environment & Release Manager →Fill your next test environment. Leave your customers out of it.
See Prune and Bloom on your own schema in a 30-minute session with our team.
docs.enov8.com/docs/datafactory/intro