Security and data

Where does my data live?

On a machine within your walls, one per organisation: there is no central server that gathers every client’s data. The main key is yours, and an encrypted backup is kept in escrow. AI models hosted online only receive what you choose to send them, with your approval; models that run on your own machine can work without sending anything. If you would rather install nothing, the same machine can be hosted for you in a data centre in Switzerland.

What is in place

One identity per person and per agent

Every agent has an accountable owner in your organisation.

Rights by role and by space

People only see the spaces and files they are entitled to.

Revoked access stops working at once

Even for someone who is already signed in.

No agent action without human approval

Outside Comoto, an agent prepares and asks; an authorised person decides. The approval covers that exact content, and the person asking cannot approve their own request. A signed record is kept.

The choice of model for each agent

Your organisation chooses each agent’s model and sets what may be passed to it.

Meotis claims no certification and no FINMA approval. This table is reviewed with every version.

Can we find the reasoning behind a decision?

Yes. Why was this asset bought? Which sources were used? Who approved it? Signed decisions are kept with their receipt in your cognitive heritage, your organisation’s memory, along with the documents that prepared them. You find what you need to explain an action and prepare for audits. In regulated environments, requirements and controls are defined with your compliance officers.

How do we govern our staff’s use of AI?

A confidential file processed in a consumer AI assistant, on a personal account? We start from how your staff really use AI and give them access that fits the company’s rules: approved data, access by role, uses explained to the teams, sensitive actions that need approval, a history you can review and an owner for every process.