Skip to content

Security and privacy

What an agent can see, what it never sees, how your customers' data is protected and how to revoke access.

The AI agent connection is built on one rule: the agent has no more access than the person who connects it. Everything else follows from that.

Scope of the connection

The agent works with your user, your role and your branch. In addition, a database-level guard rejects any write made directly with the connection token.

Inherited permissions

Every request is checked against your user:

  1. Are you an active member of the organization?
  2. Do the plan and subscription allow the feature?
  3. Did the organization turn it on?
  4. Does your role have the permission for that tool?
  5. Is the data's branch within your scope?

If a step fails, the response says so. It never returns an empty list as if there were no data: that would make you believe there are no rentals when the real issue is a missing permission.

What it can see

DataDetail
RentalsStatus, operational stage, dates, customer, line items and amounts.
Daily operationsPickups, returns and overdue rentals.
AvailabilityFree units in a range, discounting rentals and maintenance.
CustomersProfile and summary; contact masked by default.
ProductsCatalog, search by name or SKU, and utilization.
PaymentsPending collections and balances.
QuotesStatus and amounts; never the public link or signature.
MaintenanceEquipment out of service and scheduled, without costs or parts.
ReportsRevenue, rankings and business summary.

What it never sees

  • Internal notes on customers or rentals.
  • Contracts, signatures and public document links.
  • Maintenance costs, parts and work hours.
  • Data from another organization, even if the same person belongs to both.
  • Sections your role does not have permission for.

Contact details

By default, phone numbers and emails are returned masked:

Phone:   ***5678
Email:   a***@example.com
Address: [hidden]

An owner or admin can turn masking off in Settings → AI agents. If they do, the organization is responsible for informing its customers and for having a lawful basis to share that data. Turning the connection off returns the setting to “masked” on its own.

Where the data goes

This is the part worth understanding:

Leaxio does not send your data to an AI provider. The app you connect does, under your account and its own terms. Leaxio only exposes the data to that app, using the user's session and the permissions they already have in Leaxio.

Each person chooses the provider (Anthropic, OpenAI or another). It is not a Leaxio subprocessor and we cannot promise how it handles the information: that depends on its policy and on each user's plan with it. What Leaxio does control is which data it delivers and to whom.

Logging and revocation

Every request is recorded in the activity log with the tool used, the organization, the date and the outcome. Arguments and response content are not stored.

  • Access lasts 1 hour and renews itself while the app remains authorized.
  • When you revoke, the app cannot renew. A token already issued may keep working for up to 1 hour.
  • Turning the feature off for the organization cuts everyone's access.

Limits that also protect

  • 25 rows per response and a size cap: it prevents someone from extracting the whole catalog in a single request.
  • 60 requests per minute per user.
  • Date ranges of 366 days at most.
  • Internal errors are not shown to the agent: it receives a clear message without system details.

For the full terms, see the Privacy policy and the Terms and conditions.