Sensitive-data transformation
Convert any data file into any format, and keep your data out of the AI.
Describe the transformation in plain English. Data Shepherd writes a tested, reusable script that turns fixed-width, delimited, JSON, EDI, HL7, or FHIR into exactly what you need, then runs it sealed, with nothing sent to an AI model. Built for teams handling sensitive data in healthcare, finance, and operations.
Free plan available. No card required.
0001 JANE DOE 20240131 000125000 USD 0002 ACME LOGISTICS LLC 20240131 001480050 USD 0003 RIVERSIDE CLINIC 20240201 000093200 USD
| acct | name | date | amount |
|---|---|---|---|
| 0001 | Jane Doe | 2024-01-31 | 1,250.00 |
| 0002 | Acme Logistics LLC | 2024-01-31 | 14,800.50 |
Why teams switch
Your team is already pasting sensitive files into ChatGPT.
When a file is due and engineering is backed up, someone drops it into a chatbot to reshape it quickly. It works, and it is a data-exposure incident waiting to happen. Get the speed without putting customer, patient, or financial data in front of a model.
of workplace AI interactions involve sensitive data, often through unmanaged personal accounts.
Source: Cyberhaven
How it works
From a messy file to a tested, repeatable pipeline, in minutes.
No engineer, no brittle spreadsheet macros, no six-week mapping project.
Describe it and bring your file
Plain English, plus a sample or a spec / data dictionary. Have only the layout and no data to share? Build from the dictionary alone.
AI writes a transparent script
You get reviewable, versioned Python, never a black box. Inspect every line before it touches a record.
Run it anywhere, with no AI
Trigger by API, schedule, or click, or pull from SFTP / S3 / Azure. The sealed script does the work; nothing goes to a model.
The AI builds it. A sealed script runs it.
Most AI for data tools push your records through a language model on every run. We keep the model in the build phase only, so production never sends your data anywhere.
Full data-flow diagram, sub-processor list, and what-we-send-to-AI table in our Trust Center →
Use cases
If it is a file that needs reshaping, it is a job for Data Shepherd.
Legacy file modernization
Parse positional / mainframe extracts into clean, typed columns.
Trading-partner onboarding
Map purchase orders to your API schema in hours, not weeks.
Healthcare claims
Flatten and validate 837 claims for your clearinghouse or analytics.
Vendor data cleanup
Dedupe, standardize, and validate inconsistent spreadsheets at scale.
Clinical interoperability
Map HL7 v2 messages to FHIR R4, or the shape your EHR expects.
API to warehouse
Reshape nested API JSON into analytics-ready, flat rows.
Working in healthcare? There is a dedicated page for claims, HL7, and FHIR workflows →
Security and compliance
Built so your security team says yes.
No data to AI at run-time
Guaranteed by architecture
SOC 2 Type II
in progress
HIPAA BAA
Self-serve · Pro plan and up
Full audit trail
Every run and change logged
Any file in. Any format out. Your data never runs through an AI.
Transform a sample file in minutes, and hand your security team a data-flow diagram instead of a risk-acceptance form.