Use caseData privacy

Consumer privacy requests in 45 days

A former customer asks for every piece of data you hold on him. State law gives 45 days, and masking other people out eats most of them.

One question finds his documents across the repositoryThird-party data is proposed for masking, a person approvesThe case keeps a record of exactly what left

Actions on this request
The assistantDocuments found

The question returns 176 documents under his name, his two policy numbers and his old claim, each with its document class.

source · the repository, 5 document classes
The state of the copy

Masks still to approve · the AI lists where, page by page

01

Where it stalls today

The short version

Nov 3: Brian Holloway's request lands on the customer portal. His data sits in two policies, an old claim, scanned letters and emails filed on other cases. Today the privacy analyst opens each one, reads every page for other people's names, and tracks the deadline in a spreadsheet. The 45 days go into reading, page after page.

Turn the order around. One search lists everything about him, the AI proposes what to mask on each page, and the deadline lives on the case. The analyst and the privacy officer decide on every document and every mask, and the case remembers who decided what.

Request CPR-2026-0147 · Work listCase management
Day 4 of 45. Keisha's work list on Brian's request, and the note that hands the privacy officer the two calls she will not make alone.
Access requestsCase management
Every clock in one place. Each open request with its due date, sortable like any other field.

Put the work list next to the request queue and the calendar of this request shows: the deadline is set the day the request arrives, the scope is known on day two, and every page that leaves has been approved by a person. Below, Brian's request in the nine moves Keisha and Laura actually made.

02

The request

The clock starts the day the request arrives. First, know whom it concerns and where his data sits.

Step 1Intake

The request becomes a case with its own deadline

Brian Holloway fills out the privacy form on the customer portal on Nov 3. The portal calls the repository API: a consumer privacy case is created, with his message as the first document and a due date 45 days later.

Received: Nov 3Due: Dec 18Owner: Keisha Brooks
Request CPR-2026-0147Case management
The deadline is a field of the case, with a timer the team lead can see.
The 45 days are counted on the caseNobody keeps the date in a spreadsheet. If the request stalls, an escalation goes out while there is still time to act.
Step 2Workflow

Identity checked against what the company already holds

Keisha compares what Brian wrote in the form with his former policy record: date of birth, mailing address in Columbus, policy number. They match. She then records on the case every identifier the search will use.

Two email addressesPolicies HOM-55102-01 and AUT-30418-02Claim CL-21-04411
One to debate with legal: how much identity proof can you ask for before the check becomes a privacy problem of its own?
The search starts from agreed identifiersThe list is on the case, so anyone who picks the request up later knows exactly what was searched for.
Step 3Assistant

One question finds his documents

Keisha asks the assistant, in plain words, for every document that carries one of those identifiers. The question becomes a query on the repository. The answer is a table, each row naming its document class, and she exports it to Excel for the file.

The search runs with Keisha's own rights. The privacy team was given read access to the five classes concerned, and nothing outside them comes back.

Search resultsAI console
Each row opens its document. The source class is named on every line.
The size of the request is known on day twoThe team knows the volume while the clock has barely started, and can plan the review instead of discovering it.
03

The review

Finding the documents is half the work. Now each one is checked, and other people are taken out of it, page by page.

Step 4Viewer

Check that each document is really about him

Six results carry the same name, a date of birth in 1958 and an address in Dayton. Keisha opens two of them in the viewer, confirms it is another customer, and sets all six to Out of scope with one reason written on the case.

Relevance reviewCase management
Out of scope is a decision Keisha records. The table only shows why a row looks doubtful.
Another customer's file does not leave by mistakeThe doubtful rows are visible before the review starts, and each exclusion carries the name of whoever made it.
Step 5Viewer

Third-party data proposed for redaction

The 2021 claim file names the other driver, her phone number and a witness. In the viewer, the AI flags each of them as personal data about someone else and proposes a mask. Keisha accepts the masks on the driver and the witness, and keeps the body shop's name, which is business information.

Other driver's nameHer phone numberWitness's nameBody shop name: kept
Claim CL-21-04411 · Appraiser report.pdfViewer
The original stays intact. The masks apply to the copy that will be sent.
Other people are protected page by pagePrivacy law does not let one person's copy expose other people's data. Each mask is proposed with its reason and approved by a person.
Step 6Workflow

The privacy officer rules on what is held back

Two items are more than a mask: an internal email where a claims adjuster gives an opinion about Brian, and a note from the legal department on a past dispute. Keisha sends both to the privacy officer's approval queue. Laura Bennett keeps the email with the colleague's name masked, and asks the legal department before deciding on the note.

Privacy officer approvalsCase management
Each decision is written by the privacy officer, in her own words, on the item it concerns.
The hard calls have an ownerA borderline item is not settled by whoever happens to hold the file. It goes to the privacy officer, and her reason stays with it.
04

The answer

Every mask is approved. What remains is the letter, the delivery and a record a regulator can read.

Step 7Assistant

The cover letter is drafted from the company's records

A right-to-know answer covers more than a copy: the categories of data, why they are held, who received them and how long they are kept. Keisha asks the assistant for a cover letter. The draft takes each of those points from the data inventory and the retention schedule, and cites them. She rewrites the opening and checks every heading.

Response packCase management
The facts in the letter come from what the company already declares.
The letter matches the data inventoryNobody retypes the purposes or the periods from memory. The draft cites where each one comes from.
Step 8Viewer

The copy is read in his customer account

Laura approves the pack on Nov 26. The masked copy is published to Brian's account on the customer portal. He reads it in his browser and downloads it the same evening. The original files never leave the company's infrastructure, and the masks travel with the export.

Request CPR-2026-0147Case management
Every action on the copy is recorded, with its date.
The answer goes out on day 23Three weeks before the deadline, and the date is on the case for anyone who asks later.
Step 9History

The case closes with a record of what was sent

The case closes. Tom Becker, the records administrator, can show at any time which documents went out, in which version, which masks were applied, and who approved each one. If Brian writes again next year, the team starts from this list.

Register of requestsCase management
The record lists what was sent. The original documents are untouched.
The proof is ready before anyone asksWhat was sent, when, with which masks and on whose decision. A regulator gets an answer from the file.
More on thisContent governance
05

The outcomes

On any given day, Brian's request sits in one of three states. Each state tells Keisha what she does next, and who else has to look.

Ready

Every document is reviewed

In scope or out, masked or clear, each one has a decision and a name.

Then: the privacy officer signs the pack and the copy is published.
Attention required

Masks are still open

The AI lists the pages where it proposed a mask that nobody has approved yet.

Then: the analyst accepts, extends or rejects each one.
Held back

An item may be withheld

A legal note, another person's opinion. The analyst flags it with her reason.

Then: the privacy officer decides, and the reason is written on the item.
What the AI does not do

It withholds nothing, sends nothing and masks nothing on its own. Keisha approved every document that left, and Laura signed the pack, each under her own login, with her name on the case.

The AI searches, flags and drafts. The rules live in the case itself, so they still hold when requests arrive by the dozen.

06

Time back

Counted on Brian's request, answered on day 23. A company with more applications holding customer data will count higher.

1 questionto find the documents across five classes, instead of one search per application.
0 originalschanged. The masks apply to the copy that is sent; the files in the repository stay whole.
57 maskseach approved by a person, with its reason kept on the case.

Where these numbers come from. They are counted on the scenario described on this page, with example data. They are not a customer result.

Why these numbers hold

Zero wasted readingProductivityThe search and the first pass on third-party data are done before the analyst reads a page.
Zero lost deadlinesProcess reliabilityThe due date, the approvals and the copy live on the case. Nothing waits in an inbox.
Decision supportA person decidesEvery mask and every held item is a human decision, recorded with its author.
07

The people involved

Brian's request crosses three desks before his copy goes out, and each desk worries about something else.

Privacy operations

Privacy analysts

Most of your 45 days go into searching and reading, page after page.

  • One question lists every document about the person, with its class.
  • Doubtful matches show up before the review starts.
  • Third-party data is proposed for masking on each page.
  • The cover letter is drafted from the data inventory.
Risk and compliance

Privacy officers

You sign a response you never had time to read in full.

  • Borderline items reach your queue with the analyst's reason.
  • Every mask carries its author and its date.
  • The copy sent is kept exactly as it left.
  • The AI decides nothing that binds the company.
IT and records

Records administrators and CIOs

Each privacy request lands on your desk as one more extraction job.

  • The request arrives through the API, from your existing portal.
  • The search respects the rights of whoever runs it.
  • Every AI call is logged with its question, answer and model.
  • Hosted where your customer data already lives, on site or in the cloud.

How to begin

With one closed request. Take an access request your team answered by hand last year, and run the search and the redaction review on the same documents. You will see the difference on material you know.

Nothing new to install around it: the portal, the case files and the documents are the ones you run today.

Your move

Try it on one of your past requests

Bring a closed access request and a sample of its documents. We will run the search and the redaction review with your privacy team on that material.

The screens on this page are drawn to show one step at a time, with example data.

Uxopian Software helps organizations put AI to work across the systems they already run, without giving up control of their data, their infrastructure or their critical processes.