Folder-level scope
Focus the search on the files that belong to the current project.
Deep Search
Investigate project files without manually opening every document
CFAI Deep Search combines working-folder scope, local retrieval and multi-step investigation to find and connect information across a project.
No credit card required. Windows desktop app.
Multi-step investigation
Refine queries as new evidence and terminology are discovered.
Reusable knowledge
Keep useful findings in the focused project session.
01
Beyond a single semantic search
A complex question may require several searches across different file types and terms. Deep Search can iterate through the working folder, inspect relevant context and use intermediate findings to guide the next step.
The final quality still depends on file coverage, OCR quality, indexing and the model selected for synthesis.
02
Private retrieval with a clear inference boundary
Indexing and retrieval can happen locally so entire folders do not need to be uploaded as a collection.
If a cloud model is used to answer the question, relevant retrieved passages can be included in that request. Use a local model when the workflow and hardware support a fully local path.
Practical workflow
Run a focused file investigation
Choose the folder
Scope the session to the intended project directory.
Build the index
Make supported documents available for local retrieval.
Ask the question
Describe the evidence, comparison or answer you need.
Review the sources
Check the underlying files before relying on the synthesis.
Clear boundaries
Local, cloud and human control
Working-folder indexing and retrieval are designed to run locally on supported files.
Relevant retrieved passages can be sent to a selected cloud model when that inference path is used.
Users choose the folder, model and final use of the answer, and should verify important claims against source documents.
Questions
What to know before you start
Does CFAI upload my whole folder?
CFAI supports local indexing and retrieval. A cloud model may receive relevant excerpts included in a selected inference request, depending on configuration.
Which files can be searched?
Support depends on the file parser and document type. PDFs and common project documents are typical inputs, while scans may require OCR.
Can Deep Search work offline?
A fully local workflow is possible when indexing, retrieval and the selected compatible model all run locally.
CFAI AI Meta Layer
Put CFAI to work on your computer.
Create your account and experience CFAI with 10 free Cloud Credits.