Your documents stay on your machines.

Documents, tables and the search index are stored on machines you control, in that installation's own database. There is no shared database across organisations.

Answers are generated by models running on the same hardware. The models are downloaded when the system is first installed; after that, nothing is sent to an AI provider unless you deliberately configure one.

In the team setup, everyone signs in, and what each person can see follows their role, down to individual columns that hold personal data.

A single-person setup has no sign-in, so it should stay on that person's machine and not be exposed to a network.

In the team setup, actions are written to an audit log in which each entry is chained to the one before. Altering or removing an entry breaks the chain, and a built-in check shows it. The latest position of the chain is also saved daily outside the database.

The log reveals tampering; it cannot prevent someone with full control of the machine from attempting it.

Any action can be set to show a preview and wait for a person to approve it.

Connections to other systems, such as email, file storage, team chat or accounting software, are set up one by one with your own accounts. None are switched on for you.

An assistant can be shared with a named partner organisation on your network. You choose exactly who may call it, and can revoke that access at any time. This uses the Aroha protocol, an open protocol (draft specification), so the mechanism is public.

Answers sent to another organisation have common personal identifiers removed first, such as email addresses, phone numbers and tax IDs. This is best effort, not a guarantee.

We do not hold SOC 2 or ISO 27001 certification today.

If you need a security questionnaire answered, send it to us and we will answer each question plainly, including the ones where the answer is no.

Found a security issue? Follow the Aroha Labs security policy, or email contact@aroha-labs.com.