FAQ
Questions before you hand off work. GDPR, access, cost and setup.
Where your data is processed, who can see what, what it costs and how much work the setup is. If your question isn’t here, ask us in a demo.
GDPR and your data. Processed in the EU, by an EU company.
Where is our data processed?
In the European Union. Agents run in isolated microVMs in European data centers that are owned and operated by European companies, with no dependency on US cloud providers. If you’d rather keep everything in-house, zwrm also runs on your own servers. The exception is models hosted outside the EU: if an agent works with Claude or Codex through your Anthropic or OpenAI account, for example, its requests go to that provider and may leave the EU. The “EU models only” setting rules this out for your whole organization.
Who is behind zwrm?
GTN Group AG in Freiburg, Germany. As an EU-registered company it isn’t subject to the US CLOUD Act, and the contract falls under German law.
Do we get a data processing agreement?
Yes. zwrm processes your content on your behalf, as a processor under Art. 28 GDPR. A data processing agreement (DPA, in German AVV) is available on request. So is the current list of subprocessors.
What happens to our data when an agent uses an AI model?
To work on a task, the agent sends the text it needs to the model it runs on. Open-source models on zwrm are hosted in Europe. If an agent uses Claude or OpenAI, the request goes to Anthropic or OpenAI under your own account and their terms.
Will our data be used to train AI models?
zwrm uses your content only to provide the service. Requests to Claude or OpenAI fall under the terms of your own account with Anthropic or OpenAI, so check the data settings there. We’ll go through this for your setup in a demo.
How is our data protected?
It is encrypted in transit with TLS 1.3 and at rest with AES-256. Each agent runs in its own isolated microVM. Tools you connect through the zwrm gateway keep their credentials in the gateway, never on the agent’s machine.
Can we have our data deleted?
Yes. You can delete your account at any time. When the contract ends, we keep your data for 30 days and then delete it for good. Billing records are the exception: German tax law requires us to keep them for ten years.
Access and control. Each agent sees only what you connect.
What can an agent see?
Only the tools, pages and folders you connect to it. Access is set per agent, so the finance agent can work in DATEV while the sales agent never sees it.
Whose account does an agent use?
Its own. You sign the agent in to a tool once, ideally with an account set up for it, and it acts as that user. Replies and edits show the agent’s name, and it can only change what that account may change. An agent never borrows your account.
Can we take access back?
Yes, at any time. Revoke a sign-in on the agent’s page in the dashboard, switch a connection off for all agents at once, or delete the agent, which revokes all of its sign-ins.
Can an agent send emails or payments without asking?
Only if you let it. For tools that act on the outside world, such as sending email or making payments, you can require approval. The agent then stops before that step and waits until someone approves or rejects it in the dashboard.
Can we see what an agent did?
Yes. Every run is logged in the dashboard, with the tools the agent called and their results. The audit log records changes to the setup and which account made them.
Who on our team can set up agents?
That depends on their role. Members chat with the agents and connect their own accounts. Operators can also create and configure agents, connections, schedules and triggers. Admins can also invite people and manage roles.
Cost and billing. Plans from €49 a month.
What does zwrm cost?
Starter costs €49 a month and Pro €490 a month. With annual billing you save 10%: €44.10 and €441.00 a month. Enterprise is priced for your setup. All prices are in euros, plus VAT where applicable.
What’s the difference between the plans?
Every plan includes every feature. Pro runs 50 instead of 10 vCPUs at the same time and includes priority support and free onboarding. Enterprise runs on your own servers, with unlimited usage.
Can we try zwrm first?
Yes. New accounts start with a free trial, and you don’t need a credit card for it. Each person and each company can use one trial.
What do the AI models cost?
Through zwrm’s model gateway, agents use open models such as Mistral, GLM and Kimi, hosted in the EU. You pay for that usage from your prepaid zwrm credits. Agents that run on Claude or OpenAI use your own account with Anthropic or OpenAI, and the provider bills that usage to you. For every agent you can set a daily limit on runs and model spending.
How do we pay, and can we cancel?
You pay in advance for the month or the year. Plan changes, billing and cancelling are in the dashboard, and so are your invoices.
Plans per month
- Starter €49.00/month
- ProMost popular €490.00/month
- Enterprise Custom
Setup and effort. What it takes from you and your IT.
How long does setup take?
Creating an agent takes minutes in the dashboard. How long it takes to get a routine right depends on the task, and that’s what the demo is for: bring one, and we’ll show you an agent doing it. Pro includes free onboarding.
What do we need from IT?
For most tools, one sign-in per agent. For Microsoft 365, your IT registers an app in Entra ID once and grants consent if your tenant requires it. After that, people and agents sign in with their own accounts.
Do we need our own servers?
No. On Starter and Pro, agents run in the zwrm cloud in Europe. If you want zwrm in your own data center or private cloud, that’s the Enterprise plan, run by your team or by us.
What help do we get?
On Starter, you set things up yourself with our documentation and email support. Pro and Enterprise include free onboarding and priority support.
How do we start?
Book a demo and bring one recurring task and the tools it touches. Or start the free trial in the dashboard right away.
Getting started
- 01Book a demo with one recurring task
- 02Connect the tools it touches
- 03Describe the routine in plain words
How the agents work. In your tools, on a schedule, in the app.
What is a zwrm agent?
An AI agent with a name, a job and only the access you give it. It works in your tools, answers in the cowork app and runs routines on a schedule. Every run is logged in the dashboard.
Where does our team work with the agents?
In the tools you already use. Mention an agent in Notion, forward it an email in Outlook or Front, add it to a Teams channel or assign it a GitHub issue. For direct questions and files there is the cowork app.
Which tools do the agents work with?
Outlook, Teams and the rest of Microsoft 365, Notion, HubSpot, DATEV, Front, GitHub and Slack, among others. Any tool with an MCP server can be connected through the zwrm gateway, including systems your own team runs.
Which AI models run the agents?
Claude or OpenAI models through your own subscription, or open-source models hosted entirely in Europe. You choose the model for each agent.
Which languages do the agents work in?
Agents reply in the language you write to them in, such as German or English. Demos with our team are in German or English.
What if an agent gets something wrong?
You see every run in the dashboard with what the agent did. Before steps with consequences, like sending an email or a payment, the agent can wait for approval. A failed run is marked as failed with its output, and you can run it again.
Tell us what lands on your desk every week. We’ll show you an agent doing it.
A demo with our team, in English or German. Bring one recurring task and the tools it touches.