Intelligence · Module
Synthetic Data Studio
Realistic test and training data that does not expose real people
Synthetic Data Studio is planned to generate relational synthetic datasets and sandboxes that keep useful structure without exposing real people. Every release carries a privacy report and a fidelity score and needs Privacy Reviewer approval.
Useful data without real people in it
Testing, demos, trials and agent training all need realistic data, and copying production data is risky. Synthetic Data Studio, planned for W4, generates synthetic data and proves it is safe before release.
From profile to dataset
It profiles the source schema from datasets or app objects (types, distributions and relationships) without exporting raw values. Generation preserves keys and referential integrity, so accounts, contacts and deals still relate correctly. Constraints keep formats valid for Indian data: GSTIN and PAN patterns, PIN codes and localised names and addresses, with identifiers that do not match real registrations.
Proof before release
Every candidate dataset gets a privacy report: re-identification tests, distance to the closest real record and membership-inference checks. Failing tests block release. A fidelity score measures marginal and joint distribution similarity and a utility test on a declared task. Release needs a Privacy Reviewer, who cannot be the requester. Rare categories below a minimum count are suppressed, and source data must carry a purpose tag that permits the use (DPDP basis to be verified).
Where it is used
Released datasets seed tenant sandboxes for trials and training, feed Training Gym scenarios and Evaluation Studio in AI Workforce, and provide previews for Data Marketplace listings. Synthetic outputs are labelled synthetic in every surface and file. Synthetic documents for Knowledge evaluation sets are planned for W5.
How it fits
Synthetic Data Studio needs a source schema, either from a Data Fabric dataset or from app objects such as CRM accounts and deals. It works closely with AI Workforce: Training Gym uses synthetic customers and scenarios to rehearse Digital Employees before promotion, and Evaluation Studio uses synthetic golden tasks. Sales engineers can seed a trial tenant with realistic Indian data so prospects can try One3 without loading their own customers.
Features
What Synthetic Data Studio includes
Source schema profiling
Types, distributions and relationships without exporting raw values.
Relational generation
Keys and referential integrity preserved.
Privacy report
Re-identification, distance-to-closest-record and membership-inference tests; failure blocks release.
Fidelity score
Distribution similarity and a utility test on a declared task.
Release gate
Privacy Reviewer approval with segregation of duties.
Indian formats and constraints
GSTIN and PAN patterns, PIN codes, localised names and addresses.
Sandbox seeding
Synthetic data for trials and training, labelled synthetic in the UI.
Export to Training Gym and Evaluation Studio
Scenario and evaluation data for AI Workforce.
Rare category suppression
Categories below a minimum count are removed.
Workflow
Generate, test, release
- 1
Profile the source schema
- 2
Generate synthetic data
- 3
Run privacy and fidelity tests
- 4
Adjust the generator or drop columns if privacy tests fail
- 5
A Privacy Reviewer approves
- 6
Release to a sandbox, Training Gym or export
AI
AI skills and Digital Employees
Relational generation
Generates candidate datasets within the specification.
Privacy tests
Always-on tests that block release on failure.
Localised values
Generates Indian-format values that do not match real registrations.
Digital Employees that use this module
They act through the same governed Actions as people; risky steps wait for approval.
Frequently asked questions
Can synthetic data leak real records?
Privacy tests check re-identification and closeness to real records, and failing tests block release.
Who approves a release?
A Privacy Reviewer who is not the requester.
Is it labelled?
Yes. Synthetic outputs are labelled synthetic in every surface and file.
See Synthetic Data Studio in your workflow
Tell us about your team and the outcomes you want. We will walk you through the apps, the Digital Employees and the plan for your industry.