Title: Private deployment in Czechia · Help · Steerbay
URL: https://www.steerbay.com/en/docs/private-deployment/
Language: English (Česky: https://www.steerbay.com/cs/docs/private-deployment/)
Summary: Today: a dedicated company server, started after a call. Planned: local processing on hardware in Czechia.

---

# Private deployment in Czechia

Today: a dedicated company server, started after a call. Planned: local processing on hardware in Czechia.

## A dedicated server: offered today

We set up a server for a company that it shares with no one else. We start it together: it begins with a call on which we agree the setup, the scope and the price. The For companies page describes how the call works.

- The company's data, meaning files, conversations, workflows and results, is stored on our servers only, not with other services.
- Every process the agents run for the company runs on this server: the agent's browser, tasks, schedules and the work on files.
- Data is encrypted in transit and at rest.
- Strict rules forbid sending data off the server. Only what the company approves goes out.
- Only people the company approves can reach the server and its data.

## Planned, not yet offered

We plan private deployment on hardware in Czechia for companies that need to process data locally. Specialized and local models will run on that hardware too, so no external provider reads the requests. It is not offered yet and we give no date or price. The dedicated server described above is offered today.

## The server and the model are different

The working server stores files and runs tools; the AI model reads requests and writes answers. When the dedicated server uses an external provider's model, that provider reads the requests and the context they need, so a server of your own does not by itself mean fully private processing. That is why the call settles which model the server uses and what it may see, and why specialized and local models are planned too. Work on the public web always means communicating with the target website.

## What must be agreed

- The types of data and which of them may leave the environment.
- The model and its quality for your languages, documents and actual tasks.
- Separation of users and data, backups and rules for administrator access.
- Maintenance, updates, incident handling and data recovery.

We go through these on the call and agree them before the server goes live. This website does not establish certifications, a security audit, legal compliance or uptime guarantees. The AI help on this website is not part of private deployment and uses the Anthropic API.

Missing an answer? The AI help searches every guide and shows its source.