Data de-identification
Every name struck out, every change on record.
An operator marks the names, IDs and account numbers in your files and writes a copy where each one becomes a token. It logs every change and lists what it is unsure about. Your privacy team approves before the copy moves.
ISO 27001 · GDPR · Every run auditable
The change log lists the original values, so it stays with your reviewer. Only the redacted copy goes out.
Illustrative batch · Batch CL-2210
SharePoint › Claims › 2026-Q3 export
Claimant: Giulia Marchetti[PERSON_1], born 03/11/1979[DATE_OF_BIRTH_1], Via Solferino 14, Milano[ADDRESS_1]. Policy POL-7733-201[POLICY_1]. Pay to IBAN IT60 X054 2811 1010 0000 0123 456[IBAN_1].
5 changes in this file, none left open.
- 1Giulia Marchetti → [PERSON_1]Name · p.1, line 2
- 2IT60 X054 … 456 → [IBAN_1]Bank account · p.1, line 3
4 files · 11 changes · 1 item left for a person.
Enterprises running operators on Alomana
- up to 80%less time to prepare safe data
- up to 70%saved against manual methods
- up to 60%lower manual data-prep cost
Six jobs on every batch.
The work between “we can't use this data” and “here is the copy”.
- 01The filesOpens PDFs, Word, Excel, CSV and scans, reading scans with OCR
- 02The identifiersMarks names, IDs, account numbers, addresses, dates of birth and health details
- 03The tokensSwaps each one for a token, the same token for the same person in every file
- 04The doubtsReads it again and lists what still might identify someone, for a person to decide
- 05The logWrites what changed, where and why, file by file
- 06The releaseDrafts the release note and waits for a yes before the copy moves
From folder to release, in three steps.
- 1
Connect
The folder, your rules on what to remove, where the copy goes.
- 2
Redact and log
Reads each file, writes the copy and the change log.
- 3
Approve the release
You say yes. It uploads and logs it.
Know where the files go.
A model reads each file to find what identifies people. Here is where that happens.
- A private instance for enterprise
- The models your company approves
- Your data never trains AI models
- An isolated sandbox for each session
Nothing is uploaded without a yes.
- Uploads and sends wait for a person's yes
- A declined action is not retried
- Every run logged: input, output, tools, user
- Measured on
- Hours per release
- Changes reviewed per batch
- Items left for a person
- Time from request to release
- Works with
- SharePoint
- Google Drive
- Google Sheets
- Snowflake
- PostgreSQL
- Salesforce
- Jira
- Other systems via MCP
- Trusted for
- ISO 27001
- GDPR
- Never trains on your data
- Single-tenant for enterprise
A custom operator typically goes live in up to 4 weeks.
Questions, answered.
What is a de-identification operator?
An autonomous agent that prepares files for AI use. It opens each file in a batch, replaces names, IDs and account numbers with consistent tokens, writes a change log and lists what it is unsure about, then waits for your approval before the copy moves.
Is it a PII detection engine with an accuracy rate?
No. A model reads each file and marks what identifies people, and the operator shows every change and every doubt in its log. A person reviews the log and approves the release. We publish no accuracy figure; every change and every doubt is in the log for a person to judge.
Where are our files processed?
A model reads the files to find what identifies people. In an enterprise deployment that happens in your single-tenant instance, with the models your company approves, and your data is never used to train AI models. Uploading the redacted copy waits for a person's yes.
Which files and systems does it work with?
PDF, Word, Excel, CSV and scanned documents. Files come from SharePoint, Google Drive or an upload, and tables from Snowflake or PostgreSQL. Alomana has 20+ connectors, activated per deployment; other systems connect through a remote MCP server.
Can it help with the GDPR paperwork?
Yes. It drafts the release note for each batch: what was in it, what was removed, what was left for a person, who approved it, and a DPIA screening. Your privacy team reviews and signs it; the operator does not certify compliance.
Is our data private?
Alomana is ISO 27001 certified and GDPR-compliant, and your data is never used to train AI models. Enterprise deployments run single-tenant, in a dedicated instance.
Bring one batch of files.
Talk it over with sales: what the operator needs from your systems, and how it goes live.
Alomana is a secure enterprise AI platform where autonomous agents, called operators, run business processes end to end. Every run is fully auditable.