In short
A self-hosted AI workspace runs in infrastructure your organization operates. Pharen Hub, an Open Fair-Code AI workspace from Pharen IT GmbH, offers this deployment with a license from €20 alongside managed cloud and private setups. Hosting the workspace and hosting its AI models are separate decisions: an external model provider or integration may still receive data when the workspace runs on your own server.
Evaluate the full path from source documents to model requests, logs and backups. Your team needs responsibility for updates, access reviews, recovery and incidents. Self-hosting gives you control over these choices, but does not by itself establish GDPR compliance. Model connections, service providers and operating procedures must be reviewed for the intended use.
A self-hosted AI workspace runs on your own infrastructure. Tasks, documents, customer data and workflows stay in an environment you control. Pharen Hub is built as an open fair-code AI workspace and can run as managed cloud, self-hosted or private deployment. Whether external AI models or integrations process data depends on your configuration.
Many teams want AI while keeping control of their operating context. That becomes more important when the workspace contains documents, customer data, decisions, workflows and agent actions. For some teams, self-hosting is a precondition for trust.
Why does self-hosting matter for AI?
| Concern | Why it matters | Pharen Hub direction |
|---|---|---|
| Data location | AI work can include customer data, internal decisions and sensitive documents. | Support for controlled deployments instead of only one SaaS path. |
| Technical transparency | Teams need to understand how the workspace handles data and workflows. | Open fair-code foundation that can be inspected and adapted. |
| Vendor dependency | A workspace becomes hard to leave once processes grow around it. | More operating freedom through open code and deployment choice. |
| AI control | Agents need clear boundaries, permissions and review steps. | Context, approvals and traceable workflows inside the workspace. |
If a workspace only manages to-dos, hosting matters but rarely decides the purchase. Once the same workspace connects AI agents with documents, customer data and decisions, the questions get bigger: Where does the data live? Who can audit access? Which models and providers are used? How are agent steps traced? Your team must document the answers for the chosen deployment.
Managed cloud remains an option
Not every team wants to operate servers. For some teams, managed cloud is the right start. For others, a private or self-hosted setup is required because of customer promises, internal policies or compliance expectations.
A self-hosted AI workspace should offer the same operating model and capabilities: team work, knowledge, data, workflows and AI agents in one context. The deployment choice determines who operates the environment.
GDPR and data sovereignty: what self-hosting changes
Self-hosting means your workspace data stays on your own infrastructure. You decide where it is stored, who gets access and when it is deleted. That is a practical advantage for GDPR assessments. External AI models and integrations still need to be selected and reviewed for your setup.
The vendor matters too: Pharen IT GmbH is a German company based in the Leipzig region. Your contact for product, license and support is in Germany.
AI agents in Pharen Hub are designed for control as well: they work with roles and approvals. You can limit what an agent can see and do, and critical steps stay with people. Self-hosting also gives you control over the underlying workspace infrastructure.
Whether a setup meets GDPR requirements depends on the full processing arrangement. Identify the organizations handling data and their responsibilities, including infrastructure, model and support providers. The EDPB guidance on controllers and processors listed in the sources explains why those roles must be assessed for the actual arrangement.
When is a self-hosted setup the right choice?
Self-hosting or private deployment becomes relevant when:
- teams have privacy, compliance or customer requirements.
- AI workflows touch sensitive documents or customer data.
- organizations want to inspect or adapt the technical foundation.
- long-term SaaS dependency is a strategic concern.
- AI agents need visible permissions, approvals and auditability.
Decision table
| Situation | Likely setup |
|---|---|
| Small team, fast start, no special infrastructure requirement. | Managed cloud can be enough. |
| Team handles sensitive customer workflows or regulated data. | Private or self-hosted setup should be evaluated. |
| Company wants to inspect, adapt or deeply integrate the workspace. | Open fair-code and self-hosting become strategic advantages. |
Where can you host AI workflows privately?
A private setup can place the workspace and its data in infrastructure your organization controls. For Pharen Hub, discuss whether that means your own environment or a private deployment operated under an agreed service arrangement. The workspace location alone does not establish where an AI request is processed.
Check each connection separately: the model endpoint, mail and calendar integrations, document processing, logs and backups. If a workflow sends document text to an external model, that text leaves the workspace environment even though the application is self-hosted. Document the selected endpoints and test the data path before using sensitive records.
Operating a self-hosted AI workspace
A model server provides model inference. A workspace also needs to manage the records, people and actions around it. When comparing a custom stack with Pharen Hub, evaluate the actual components and configuration; roles, document search and audit features vary by product and version.
Use these checks during a deployment review:
- Assign an owner for application, database and model updates, with a rollback procedure.
- Test restoring the workspace and its files from a backup.
- Check access with an ordinary user account and an agent account, including a document they should not see.
- Define where a workflow stops when the model or a connected service is unavailable.
- Agree who handles incidents and which support hours are covered.
Include hosting, model usage and operating time in the cost comparison. A software license alone does not cover those costs. Pharen’s deployment and security information is the starting point for checking a proposed setup; the agent guide explains the role of approvals.
Frequently asked questions
What does a self-hosted AI workspace cost with Pharen Hub?
The Pharen Hub self-hosted license starts from €20. The managed cloud has a free plan, plus Starter at €25, Team at €79 and Pro at €199 per month. Details are on the pricing page.
Is a self-hosted AI workspace GDPR-friendly?
With self-hosting, workspace data stays on your own infrastructure, and you decide where it is stored, who has access and when it is deleted. External AI models or integrations may still process data depending on your configuration, so the GDPR assessment always depends on the complete setup.
What is the difference between self-hosted and managed cloud?
For Pharen Hub the product direction is the same: one workspace approach with workflows, knowledge and agents. The difference is operations. In the managed cloud, Pharen handles hosting, updates and maintenance. With self-hosting, your IT operates the environment and carries that responsibility.
Can we start in the cloud and switch later?
Many teams start in the managed cloud because it is faster, and evaluate self-hosting once customer, privacy or IT requirements become concrete. If that is your plan, talk to us early so the requirements are considered from the start.
Can we run Pharen Hub with our own or self-hosted models?
Yes. Depending on the setup, Pharen Hub can work with selected, customer-owned, self-hosted or OpenAI-compatible models. Which models you connect is decided in your configuration; details are on the security page.
Related guides
- From tool stack to operating system for work
- A workspace where people, knowledge and agents share the same context
- AI agents become useful when they are embedded in real work
- A self-hosted Notion alternative: what Pharen Hub does differently
Next step
If you want to check whether Pharen Hub fits your stack, talk to us about your workflow or start with the cost comparison.
