Skip to content

Legal & safety

Your clients' data is the job.

Presentable is built by a Danish company on EU infrastructure, and confidentiality is a design constraint, not a policy afterthought. Here is how it holds together — plainly.

Model & keys

Who runs the AI

The agent runs on Anthropic's Claude today

One provider, chosen for quality — changes come with notice under the terms.

Bring your own key

Coming soon

Run Presentable on your organisation's own API key.

Infrastructure

Where your work lives

EU-hosted, end to end

Hetzner infrastructure and Scaleway services. Data leaves the EU only for the AI API, under SCCs with a transfer impact assessment.

Row-level security on conversations

Every chat query is pinned to the signed-in user at the database layer.

AI-written code runs in a sandbox

Fully isolated from the rest of the platform, with no internet access and hard limits on time and resources.

Data handling

What the AI never gets

Never trained on your content

You own your inputs and outputs, and our AI providers are bound not to train on them.

Raw files stay out of the model

Spreadsheets are staged into the sandbox — the model sees a small preview, not your file.

No third-party analytics

No trackers, pixels or ad-tech anywhere. Operational telemetry is self-hosted in the EU.

GDPR

Built to the GDPR

A Danish controller, documented properly

Alva AI ApS, Aarhus. Named subprocessors with locations, SCCs with transfer impact assessments, a retention schedule and 72-hour breach notification — all in the privacy policy.

The documents

Read the fine print

Questions about security or procurement?

We answer them ourselves — no ticket queue.

support@presentable.so