Book the Scan
AI/Boundaries and labels

Boundaries and labels

Before anyone switches AI on, we decide what it can see. Payroll, HR and board files stay private, and the boundary is tested with the questions you hope it can’t answer.

Plan your AI rollout
PCPaige Tab, Carl ShiftTeams chat
  1. CarlLiam asked it what everyone earns.
  2. PaigeIt wouldn’t tell him, right?
  3. CarlIt made him a table.
What we do

How it works.

  1. 01Find what is open

    Everything AI could reach today, through every way it is shared.

  2. 02Label what matters

    A small set of labels that keep AI out of private files.

  3. 03Close the gaps

    Sharing fixed where it was too wide.

  4. 04Test it

    The questions you hope it can’t answer, asked before rollout.

See it working

Ask what an employee might ask. See what AI finds today, and after the boundary is set.

Sample data. Nothing you enter leaves this page.

Open it on its own page

Examples

What the work looks like.

Sample document · made-up companySample boundary report

What AI can and can’t see.

Open the sample Sample document · made-up companySample label set

Six labels, applied automatically.

Open the sample Case study · client unnamedA manufacturer finds out what Claude would see before connecting it.

A metadata-only scan of about 240,000 files, eight maturity scores, a boundary design for Claude and a fix list in five workstreams.

Read the case study Field NotesAsk your AI what it can see before your team does.

Files that were safe because nobody could find them become answers in a chat. Ask the questions you hope it cannot answer, before anyone else does.

Read the note
Questions

What people ask.

Is this only for Copilot?

No. The same boundary protects you with Claude, ChatGPT or any tool connected to your files.

How long does it take?

About two weeks for the boundary check, then rollout.

Plan your AI rollout.

Thirty minutes on a call. We tell you how we would start. Not sure yet? The Scan is free.