1. Data processing depends on how AI is connected
Before using AI features, check where the selected model runs and which services it connects to.
SetFN is primarily an on-premises system running on the PC or server where you install it. When you use local AI, data is processed using that environment's resources, including its CPU, GPU and memory.
SetFN does not send your files or work data to external AI by default. External AI connections must be selected and configured by an Administrator or user. If you do not configure and use a connection, data is not sent to external AI through that SetFN connection.
If you choose an external AI connection, your request and authorized material may be sent to that AI service. Even when using a connection configured by your organization's Administrator, check whether processing takes place locally or at an external service.
| Usage | Where AI processing takes place | Scope of data transfer |
|---|---|---|
| Local AI | The installed PC or server, or an internal AI server designated by the Administrator | When configured without external connections, AI processing data stays within that operating environment. |
| External AI service | The external AI service you select | Requests and data needed for the task may be transferred according to connection settings and permissions. |
| AI client connected through MCP | Depends on where the model used by the connected client runs | Results from authorized tools and retrieved material are sent to the client. If it uses an external model, they may also be sent to that service. |
| Desktop app AI integration | Depends on where the model used by the installed AI program runs | If you enable the integration, task requests and related material may be sent to that program and its connected services. |
2. Local AI uses your installation's resources
To use local AI, install a supported runtime, such as Ollama, and the model you intend to use, then configure SetFN to connect to that runtime. Performance and the models you can run depend on your CPU, GPU, memory and the model's requirements.
When you run a local model without external AI or external tool connections, the files, questions and responses used for AI analysis are not sent to an external AI service. If you designate a separate internal AI server, data is sent to that server for processing.
Even if a runtime is installed locally, selecting a cloud model, external search or external tools may cause external communications. Check the actual model, connection address and external feature settings, rather than relying on the runtime's name. Some runtimes, including Ollama, offer both local models and cloud features.
Communications required to download runtimes or models, install updates or verify licenses are separate from sending analysis material externally. These guidelines do not mean that all internet communications by the product are blocked.
3. External AI is selected and configured by you
To use external AI, an Administrator or user must configure the service, account or API connection details, and access permissions. SetFN provides the means to connect to external AI; its default configuration does not arbitrarily send work material to an external AI service.
Once you configure and use a connection, data needed to process the request may be sent to the external service. This can include questions, selected file contents, search results, related records and tool results. Even if the original file is not attached, extracted text, summaries or metadata may be transmitted.
Retention periods, use for training, fees and terms of use depend on the external service and account settings. SetFN does not determine that service's processing policies. Review the service's information before connecting.
4. Connect Workspace tools through the MCP server
SetFN provides MCP (Model Context Protocol) server functionality so that AI clients can call SetFN Workspace tools. Administrators configure the MCP endpoint and the scope of external client connections and access.
When an MCP connection is authorized, the connected AI client can retrieve Workspace material or request actions within the permitted tools and permissions. The client can read file contents, search results and work records returned by tools. If it is also authorized to use tools that make changes, it can perform actions within that scope.
MCP is a way to connect tools, not an AI model itself. Where data is processed depends on whether the connected client uses a local model or an external AI service. A client using an external service may send retrieved material and tool results to that service.
Providing MCP functionality does not automatically grant access to external AI. Unless a connection is authorized and configured, external AI is not connected to retrieve material through that route. Check the client, account, permitted tools and scope of material when configuring access.
5. Integrate with AI programs installed on your desktop
In supported environments, the SetFN desktop app can connect to AI programs such as Claude or Codex installed on the same desktop. Availability and the scope of integration depend on the installed programs' versions and settings.
Installing an AI program on your desktop does not automatically transfer SetFN material to it. Task requests and related material may be sent to that program when an Administrator or user chooses the integration and configures the necessary connections and permissions.
Also, an AI program installed on your desktop does not necessarily process AI requests locally. If it uses an external model, material sent to it may be processed by an external AI service. Check the program's model provider and data processing settings.
If you do not select and configure the integration, SetFN does not send material through that integration. File permissions you separately grant to an external program, or material you manually copy or attach, are independent of SetFN's integration settings.
6. Check before use
- Check whether the model you are using is local or provided by an external service.
- Check whether the connection address points to your device, an internal server or an external service.
- Review the material, tools and actions authorized for MCP and desktop AI integrations.
- When using external AI, review the service's retention and training settings and its terms of use.
- If you do not want to use external AI, disable external service connections and the related tool integrations.
Disconnecting a service or revoking permission does not automatically delete data previously sent to it. Retention and deletion of data already transferred are governed by that service's policies and settings.
These guidelines explain the scope of data processing and connections associated with AI features. For details of information processing during product operation, also refer to each product's Privacy Policy.