AI agent for data architects
Sensitive Data Discovery Agent
A verified map of where sensitive data lives, with mismatches fixed or accepted
What it does
Personal data hides in free-text columns and old tables that nobody remembers. This agent scans schemas and samples values for personal, health and payment data. It classifies each column and checks the result against what the data catalog says. Where the two disagree, it lists a mismatch and proposes masking or tighter access. It reruns the sample on a different slice of rows to confirm it was not a fluke before raising a finding. The data owner approves each change. Edge case: a notes column has a card number in only three rows out of a million, so the agent flags the column as high risk and explains why.
How it works
Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.
Read the steps as a list
- Quarterly data scan
- Read schemas and sample values from each table
- Classify columns by the sensitive data they hold
- Compare the classification with the data catalog
- Does a second sample from other rows confirm each finding?If not: take a larger or different sample and recheck. Back to step 2.
- List mismatches and proposed masking or access changes
- Data owner approves each changeThe agent waits here for your OK.
- Apply the approved masking or access changes
- Does a rescan show the data masked or access limited?If not: repeat the change and inform the owner. Back to step 2.
- Data map and findings report
How it decides
A column is sensitive when sampled values match patterns such as card numbers or health terms. A single confirmed hit is enough to raise it.
- Match values to patterns for cards, IDs and health terms
- Raise a column on one confirmed hit
- Treat free text columns as high risk
- Update the catalog after each approved change
Make it yours
Every agent is a starting point. You choose these settings for your own situation.
- Sample size
- Patterns and categories
- Scan frequency
- Systems in scope
What keeps you in control
It always asks you first
- Data owner approves each masking or access change
Hard limits
- Never copies sampled values into reports
- Never changes data without approval
It stops when
- Done: mismatches are resolved or accepted
- Stop: access to schemas or samples is not granted
Set it up
We guide you through the set-up, step by step
Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.
- One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
- The agent then walks you through connecting your own data, one source at a time
- A downloadable copy with the flow chart, the rules and the full guide
