Local where supported
Use local transcription, retrieval and compatible local language models.
Privacy Architecture
Know what stays local and what uses a selected cloud service
CFAI is a hybrid desktop system. The exact data path depends on the mode, model and provider you choose, so privacy should be explained feature by feature.
No credit card required. Windows desktop app.
Cloud by selected feature
Metered providers are invoked only by workflows configured to use them.
Context remains scoped
Choose the session, files and visible content relevant to the task.
01
A feature-by-feature privacy model
Audio transcription, document indexing, model inference, screen vision and computer actions are different operations. They should not be described with one blanket local or cloud label.
Before using sensitive information, verify the selected model provider, the enabled tools and the context attached to the current session.
02
Local processing is a capability, not a slogan
CFAI supports local AI through compatible runtimes such as Ollama and LM Studio, along with local file retrieval and transcription workflows.
Some advanced features may require or perform better with cloud inference. When that path is selected, the relevant request data is processed according to the chosen provider and configuration.
Practical workflow
Check the data path before sensitive work
Identify the feature
Determine whether the task uses transcription, retrieval, vision, inference or action tools.
Check the provider
Confirm whether the selected model is local or cloud-based.
Scope the context
Attach only the screen, files and folders needed for the result.
Review the action
Confirm consequential operations before they affect the system.
Clear boundaries
Local, cloud and human control
Compatible local models, transcription and retrieval can keep selected processing on the Windows computer.
Cloud inference receives the information needed for the selected request and consumes Cloud Credits when the service is metered.
Provider selection, scoped context and approval gates give the user practical control over each workflow.
Questions
What to know before you start
Is every CFAI feature fully local?
No. CFAI is hybrid. Some workflows can be local, while selected vision, reasoning or other services can use cloud providers.
Are my full folders uploaded for document search?
CFAI supports local indexing and retrieval. A selected inference provider may receive relevant context used to answer a request, depending on configuration.
Does local AI guarantee privacy?
Local processing can reduce data transfer, but security also depends on device access, model configuration, file permissions and operational practices.
CFAI AI Meta Layer
Put CFAI to work on your computer.
Create your account and experience CFAI with 10 free Cloud Credits.