Security and privacy

A CPA office holds its clients' most sensitive information. Here is how we protect it, with no marketing fluff.

Separation between offices

Every office is separated at the database level: every query is filtered by office, even if there is a bug in the screen's code
Permissions by role, by client, and by team. The restriction is enforced in the database too
A client in the client portal sees only their own documents and reports

Two-factor authentication (a code from an authenticator app) is mandatory for the office owner
An alert email on sign-in from a new device
A blocked employee is signed out immediately from all devices

Documents, ID and company numbers, and connection keys are stored encrypted
Traffic is encrypted (HTTPS) on every page
Servers and database are in Europe (Frankfurt)
Credit card numbers are not stored with us. Only a token from the payment processor

Activity log: who changed what and when, including actions the system suggested and a person approved
Our support access is view-only, time-limited, with a documented reason. The office sees a banner while it is active

In a demo office, the AI works on fictional data.
On a real office's data, the AI works only after the office owner turns it on in the settings screen, after seeing exactly what is sent, where to, and what is never sent. Consent is recorded with a date and a name, and can be withdrawn at any time.
ID and company numbers, phone numbers, emails, and account numbers are masked before anything is sent. Files and documents are not sent: they are read on our server.
The data is processed by Anthropic in the US. Under their commercial terms (section B), Anthropic may not train models on this data.
The AI doesn't change data. It suggests, and a person approves. Every number in an answer is checked against the data.

The office owns its clients' information, and we process it solely on the office's behalf. Full details are in the privacy policy and terms of use.