Join the waitlist

Platform

One engine, everywhere data goes

Ligurian uses one classification engine for data at rest and data in motion. What it labels in SharePoint is the same thing it recognises in an email, a browser upload or an AI prompt.

Know

Data security posture

  • Inventory of sensitive data across SharePoint and OneDrive, with Exchange mail and Teams chats and channels as well
  • Who can reach it: internal, external, anonymous links and sharing with everyone
  • Findings routed to the site owner, with a nudge to fix them
  • Duplicate copies of the same sensitive file, wherever they've spread

Protect

Data loss prevention

  • Browser: pastes, uploads, drag and drop and form submits in Chrome, with coaching on AI sites. No proxy
  • Windows: pastes into supported apps are coached or blocked; printing is detected and coached
  • USB and network shares: caught just after the copy, quarantined on the device and removed, restorable for 72 hours. Stopping the copy before it happens comes in a later update
  • Email: a check in Outlook with a header your mail rules can act on, coming next
  • Not covered in the first release: uploads through personal sync apps such as Dropbox

Govern AI

AI and agent visibility

  • What Microsoft 365 Copilot can reach, and the sharing that lets it
  • What people paste or upload into ChatGPT, Claude and other AI tools
  • Which apps, service accounts and agents hold access to your data, and what they actually use
  • Coaching at the moment someone shares sensitive data with an AI tool

Watch

Activity and access

  • Bulk downloads, deletes and reads that don't fit the person or account
  • Logins from unusual places and repeated failures
  • Access that doesn't match someone's role
  • Movement of labelled data over time, so you can see where it went

The engine

Cheap checks first, expensive ones only when needed

Reading everything with an AI model is slow and costly, and it means sending more data further. Ligurian works through a funnel instead. Each stage only passes on what it can't settle.

The same engine runs in our cloud, on Windows laptops and, for cloud workloads, inside your own account.

  1. Already seen

    Content classified once is recognised again by a one-way fingerprint, so a labelled file is known instantly when it's pasted or uploaded.

  2. Rules and validators

    Patterns, checksums, keywords and column names. Settles most content in milliseconds.

  3. Small models

    Trained classifiers for the cases rules can't settle, run only on the candidates.

  4. Large models

    Reserved for the rare ambiguous remainder, inside our Australian hosting.

Find it, then fix it

Fixing is part of the product

Plenty of tools find sensitive data and leave you with a list. Ligurian is built to close the loop, starting with the first release.

Breach scoping

Pick a mailbox, site or user and a time window, and see what sensitive data was exposed, with the 72-hour notification clock running.

Fixes you approve

Signed scripts the customer runs to close oversharing. Ligurian stays read-only and checks that each fix took.

Reports that land

A personal information register, monthly reports and quarterly review packs, plus a monthly statement that no content was kept.

Native controls

Enforced by the platforms you already trust

Ligurian writes labels and policies into Microsoft 365 and your cloud's own access controls, so protection holds even where Ligurian isn't in the path. There's no proxy to route traffic through.

Changes are checked with the platform's own simulator where there is one, and shown as a preview you approve. Controls start in monitor mode, report what they would have blocked, and move to enforce when you say so.

What comes when

Honest about the order we build in

Ligurian is in development. This is the order we're building in, so you know what to expect from the first release.

First release

  • SharePoint and OneDrive, with Exchange mail and Teams
  • Sharing and exposure report
  • Windows agent and Chrome extension
  • Coaching on AI sites
  • Copilot exposure report
  • Breach-impact scoping with a 72-hour clock
  • Personal information register export
  • Fix scripts your customer runs
  • Customer reports and one-way PSA tickets

Next

  • Enforcement through Microsoft 365's own controls, and labels written back
  • Outlook add-in
  • ChatGPT and Claude
  • Xero
  • RMM packages and two-way PSA
  • Register of AI agents and what they can reach
  • App and service account access review
  • Agent behaviour indicators

Later

  • Azure storage and Microsoft Fabric, classified as data changes
  • AWS S3 and databases
  • AI pipelines and gateways
  • Google Workspace
  • Data warehouses
  • macOS
  • Behaviour analytics for people, opt-in
  • Google Cloud

Be first in the hive

We're building Ligurian with a small group of MSPs.

Join the waitlist