The detection runs on its own
The automation reads the batch of documents and finds the personal information: names, addresses, numbers, any identifying detail, before the officer even opens the file.
Use case · Automation
Redaction that stays consistent, document after document.
An access-to-information request comes in, and someone has to find and redact personal information across dozens of pages, consistently, within the legal deadline. Often the same person who already handles everything else. The automation finds and proposes, the access-to-information officer decides.
Where it starts
Before every request, the same race against the deadline.
An access-to-information request lands on the clerk's desk, and someone has to comb through a batch of documents, sometimes dozens of pages, to find every piece of personal information before it goes out: names, addresses, numbers, any identifying detail. All within the deadline set by Quebec's access-to-information law, and often by the same person who already drafts the resolutions, the minutes and the public notices.
- Finding personal information page after page, with no checklist or workflow.
- Staying consistent: a name redacted on page 3 has to be redacted on page 40 too.
- A fixed legal deadline, no matter how large the batch of documents.
- Doing it on top of everything else: resolutions, minutes, public notices, without losing a week to it.
What we built
A proposed redaction, consistent, ready to validate.
Built on request for clerk's offices, on a bank of hours, integrated with Munia.
The redaction stays consistent
A detail redacted on page 3 is redacted on page 40 too, across the whole batch. A checklist travels with each file, to confirm what was found and what was removed.
The officer decides
The document arrives redacted, ready to review. The access-to-information officer validates, adjusts where needed, and keeps the final decision on what goes out. Nothing is sent without that validation.
How it rolls out
From the workshop to going live.
- 1
The workshop, a half-day
Free, on your real access requests. We look at the volume, the document types, and where manual redaction eats the most time.
- 2
Development, in weeks
We configure the detection for your document types and validate with you, file by file, until the detection is reliable.
- 3
Integration with your Munia
The automation lives in the tool your team already uses. Training for the access-to-information officer is included.
- 4
Follow-up
We adjust with you after the first real requests. Connections to other systems get added on request.
Who it's for
Whoever answers access-to-information requests.
Clerk's office
The person who receives the request no longer has to find everything alone, page by page, before it goes out.
See the clerk's office pageGeneral management
The file goes out within the legal deadline, with a checklist to back it up, without pulling in the whole team.
See the general management pageRegional municipalities (MRC)
Intermunicipal boards, housing offices, municipal-sector organizations: the same automation, sized to your request volume.
See the MRC pageQuestions
What we get asked about this case.
No. It finds personal information and proposes a redaction; the access-to-information officer decides, adjusts and approves before anything is sent. The decision stays human.
They are processed on servers in Quebec, under Quebec and Canadian jurisdiction, and inference runs under a zero-retention contract. Your documents are never used to train a model.
Yes, as long as the text is legible after scanning. A document too degraded or handwritten gets validated at the workshop, before development starts.
Let's talk
One demo. 20 minutes. No commitment.
Tell us about your municipality and one pain point you'd like to solve. We'll show you how Munia (and Carange around it) handles it. Reply within 48 hours.