Start freeBook a call

Security

Safe by design.
Here is how.

Everything your people build runs on your own server in an EU data centre, behind company sign-in, with roles, approvals, spend limits and a record of every action. This page explains how it works, what the AI sees, and how we test.

Where it runs

Your own server, separate from every other customer.

Each company gets its own isolated server in an EU data centre, a dedicated server only your company uses, or an installation in your own data centre or cloud. Your apps, data, files and agents run there, not in a database shared with other companies.

Encrypted connectionsEvery connection to your server uses TLS encryption.
Keys kept lockedPasswords and keys for the services you connect are stored encrypted (AES-256) and are never shown back in full.
Company sign-inPeople sign in with their company accounts. Single sign-on is available for larger companies.

How apps and agents are contained

Every app runs in a sandbox. It can only do what you allow.

Apps run in a confined frame and reach your data only through the permissions you grant them, one by one. Code that agents write and run is executed in a separate sandbox, away from the platform itself.

RolesWho may build, who may approve and who may only use an app is set for each person.
ApprovalsAgents ask a person before any step that needs a human decision, such as sending invoices to customers.
Spend limitsEvery app has its own AI spending limit, so nothing runs up a bill on its own.
Company rulesStanding rules set by your admin that every agent and every builder must follow.
Every action on recordWhat every person and every agent did, and when, kept in an audit log you can review.

What the AI sees

You choose which AI sees what.

To write an answer, the AI model receives the part of the conversation and the documents that task needs. Your apps, data and agents stay on your server; you choose where the AI requests go.

AI from usIncluded in your plan and provided by established AI model providers under data-processing agreements. The current providers are named in the sub-processor list published in every Digital Worker.
Your own AI accountConnect your company's own AI account, and the requests go under your own agreement with that provider.
Private AI modelsFor strict data rules, AI models can run on your own server, on request, so the requests never leave it.
Only what you allowYou decide which documents and systems each agent may use.

How we test

Tested before every release.

Every change to the platform runs through our automated test suite. Before a release ships, we install it on separate test servers and attack it with open-source security tools.

This is our own testing, not an independent audit.

Vulnerability scansKnown weaknesses and risky settings, checked on a fresh installation of the release.
Encryption checksThe TLS set-up of every address the release serves.
Load and slow connectionsHow the release holds up under heavy traffic and deliberately slow clients.
Hostile outside serversHow the platform behaves when an outside server or an AI provider answers with something malicious.

If something goes wrong

If something goes wrong, you hear it from us.

If a security incident affects your data, we tell you without undue delay, as the GDPR requires of us as your data processor: what happened, what it affects and what we are doing about it. Our Data Processing Agreement sets out the details.

Found a security problem? Write to security@apexacademy.hr; we confirm receipt within one business day.

The documents behind itOur Terms, Data Processing Agreement, Service Level Agreement, Acceptable Use Policy, sub-processor list and EU AI Act statement are part of every contract and published in every Digital Worker. Ask, and we send them before you sign.
Read them nowThe Terms and the privacy notice for this site are online.