Obfuscation Portal

Detect concepts and generate semantic replacements

How It Works
1
Upload
Word, Excel, Text, PDF files
→
2
Detect
AI identifies sensitive concepts
→
3
Obfuscate
Replace with semantic equivalents
→
4
Unmask
Restore original terms in derived content

Upload confidential documents and let AI detect sensitive concepts (names, companies, projects, etc.). Selectively obfuscate these concepts while preserving semantic integrity—ideal for downstream AI analysis. When AI-generated content references obfuscated terms, unmask them back to originals automatically.

Known Limitation: Derived content may contain the same concept text as an obfuscated term (especially ubiquitous words) and may be incorrectly back-converted. Final human review is advised.

1. Select Project

2. Run Detection

Usage: Flag common words or phrases that should be considered sensitive in this project's context.
Caution: Common words flagged here may cause false positives during unmask. See known limitation.
Advanced Settings ⓘ
loading...
Detection Prompt ⓘ 

                

                
            

        

        

            

            

                
                
            

        

        
    


    
    

        

            
3. Review & Select

            
        


        

            
            

                

                    
                

                

                    
Select project first

                

                

                    
                    0 of 0
                    
                

            


            
            

                

                    -
                    (0 occurrences)
                


                

                    
Select replacement:

                    

                        Select a concept to see options
                    

                


                


                

                    
Occurrences:

                    

                        
Select a concept to see occurrences

                    

                

            

        

    


    
    

        
4. Apply

        

            
            

            
        

        
        

            

                Anonymized Output:
                
            

            

                No output files yet.
            

        

    


    
    

        
5. Unmask

        
Upload files containing anonymized content to restore original terms.

        

            Note: Ubiquitous terms may be incorrectly back-converted. See known limitation above.
        


        
        

            
or drag and drop

            
            
        


        
        

            
            
        

        


        
        

            
            
            

            
        


        
        

            
            
            

            
        

        


        
    


    
    

        
6. Image Analysis

        
Analyze diagrams, flowcharts, or screenshots using the vision LLM.


        

            Vision LLM: Checking...
        


        
        

            
Drop image here or click to upload (PNG, JPG, WEBP)