Hissa Redaction Service

Your name

File (.txt .csv .md .log .xlsx)

Header rows — click a row to mark it as a header; headers give the AI column context Loading preview…

Excel files (.xlsx) preserve layout and formatting. Charts and pivot tables may not be retained.

Redacted file (.txt .csv .md .log .xlsx)

Use a job on this server Upload mapping CSV

Job ID

Mapping CSV

What is the Hissa Redaction Service?

This tool scans your files for personally identifiable information (PII) and replaces each detected piece of data with a short, opaque placeholder such as [PERSON_1] or [EMAIL_2]. The result is a redacted copy you can safely share — for example, with an external AI tool or a third-party vendor — without exposing real names, addresses, or credentials.

A mapping CSV is saved alongside the redacted file. If you later need the original values back, upload the redacted file together with the mapping (or the job ID) and the service will restore the original text exactly.

Privacy first: your uploaded files are stored only as long as needed. Use the Delete button in the jobs table to remove them immediately once you have downloaded the outputs.

Redacting a file

Enter your name in the Your name field. This is used to tag the job so you can find it later in the table; it is not validated.
Click Choose file and select any supported file: .txt .csv .md .log .xlsx (maximum 50 MB).
If you upload an Excel (.xlsx) file, a preview of each sheet appears automatically. Click any row to mark it as a header row. Header rows are excluded from redaction — this is useful for column names that happen to look like names or emails. Row 1 is selected as a header by default.
Click Submit. The job is queued and the table below refreshes automatically every two seconds until it completes.
When the status shows done, use the Downloads column to save:
- redacted — the file with all PII replaced by placeholders.
- mapping — a CSV that maps every placeholder back to its original value. Keep this file safe.

Warning indicator: if the job has a ⚠ badge next to the "view" link, the model detected one or more patterns that may be suspicious (e.g. tokens that look like secrets or API keys). Open the view to inspect them before sharing the file.

Viewing results

Click view in the jobs table to open a side-by-side comparison of the original and redacted text. Each placeholder is highlighted and you can see exactly what was detected and replaced.

From the view page you can also edit the mapping: remove a placeholder if the model incorrectly flagged something, or add one for a piece of PII it missed. Saving edits regenerates the redacted file and mapping CSV immediately.

Restoring a redacted file

Use the Restore a redacted file tab to reverse the redaction — this replaces every placeholder with its original value.

Switch to the Restore a redacted file tab.
Choose the redacted file (the one with placeholders like [PERSON_1]).
Provide the mapping in one of two ways:
- Use a job on this server — paste or type the job ID, or click restore in the jobs table to fill it in automatically.
- Upload mapping CSV — if you downloaded the mapping CSV and want to restore the file locally or on a different machine.
Click Restore. The restored file downloads automatically.

If the status shows "X placeholder(s) had no match", those tokens were in the redacted file but not in the mapping — they will be left as-is. This can happen if the redacted file was edited after download.

What PII is detected?

The model uses a fine-tuned transformer trained specifically for privacy detection. It recognises the following entity types:

👤

Person

Names of individuals

📧

Email address

Personal email addresses

📞

Phone number

Mobile and landline numbers

🏠

Address

Physical / postal addresses

📅

Date

Personal dates (e.g. DOB)

🔗

URL

Personal / private URLs

🏦

Account number

Bank / financial account IDs

🔑

Secret

API keys, passwords, tokens

The model runs entirely on your server — no data is sent to any external service during redaction.

Jobs table reference

Column	What it means
ID	First 8 characters of the job's UUID. Use the full ID when restoring via the form.
User	The name you entered when submitting. Only jobs matching your current name are shown.
File	Original filename of the uploaded file.
Status	⏳ queued — waiting to run. 🔄 processing — actively redacting, shows % progress. ✅ done — complete; outputs are ready to download. ❌ failed — something went wrong; hover the icon for the error.
Spans	Number of PII spans detected and redacted.
Submitted	Date the job was created (IST timezone).
Duration	Time taken from start to finish of the redaction job.
View	Opens a side-by-side diff of original vs. redacted output. A ⚠ badge means suspicious patterns were found.
Downloads	redacted — the sanitised file. mapping — the placeholder-to-original CSV.
Restore	Pre-fills the Restore form with this job's ID so you can quickly reverse the redaction.
Delete	Permanently removes the uploaded and output files from the server. The job record remains but files cannot be recovered.

Limits & data retention

Maximum file size: 50 MB per upload.
Supported formats: .txt, .csv, .md, .log, .xlsx
Excel charts and pivot tables may not be retained in the redacted output — only cell values are processed.
File retention: uploaded files and outputs are stored on the server until you delete them or the automatic cleanup period expires. Download your outputs and delete the job once you no longer need them.
Mapping CSV: keep a copy of the mapping file if you may need to restore the original later. If the job is deleted, the mapping is gone.

Jobs

ID	User	File	Status	Spans	Submitted	Duration	View	Downloads	Restore	Delete
Loading…