Local AI meeting notes · v4

Upload a recording.
Get a proper breakdown.

Faster Whisper + Ollama, fully local. All uploaded files and generated outputs are automatically deleted after 24 hours.

Fully local Auto model pull 24h auto-purge Transcript + notes 3050-friendly

Service status

ServiceChecking…
Default model—
Whisper—
File retention—
Available models—

New run

Create a job

Transcript options
Good defaults for a 3050: Use gemma3:4b, Whisper small, depth standard, length medium.

Queue

Jobs

Output

Job result

No job selected.

Summary


          

          

            

              
Transcript

              
            

            

          

        

      


    


    
    


      
      

        

          

            
Floor plan extractor

            
Upload a PDF.
Get the codes in Excel.

            
Extracts codes like B1-MEP02A/B from floor plan PDFs. Supports multi-letter suffixes, odds/evens split, and per-floor split.

          

          

            Handles line wraps
            Per-floor sheets
            Odds / evens
            Up to 100 MB
            Scanned PDF / OCR
          

        


        

          
PDF info

          
FileNone selected

          
Pages—

          
Page 1 codes—

          
Sample codesUpload a PDF to preview

        

      


      
      


        
        

          

            

              
Extract

              
Settings

            

          


          


            

            

            

              Output splits
              
              
              
            


            

            

              Code pattern:
              Floor-Type##Suffix — e.g. B1-MEP02A/B, 01-MEP15A/B/C/D. Each letter becomes its own row.
            


            
          

        


        
        

          

            

              
Activity

              
Recent extractions

            

            
          

          

            
No extractions yet.

          

        

      


    


    
    


      
      

        

          

            
Cable label vetter

            
Upload the data.
Get it validated.

            
Deterministic checks for pivot counts, formula integrity, duplicates, cabinet naming, cable sequencing, port ranges & more. Optional AI narrative summary via Ollama.

          

          

            Pivot counts
            Formula checks
            Duplicate detection
            RIGHT() truncation
            Cabinet naming
            Cable sequencing
            Port ranges
            AI summary
          

        


        

          
Dataset info

          
FileNone selected

          
Rows—

          
Columns mapped—

          
Missing requiredUpload a file to check

        

      


      
      


        
        

          

            

              
Validate

              
Settings

            

          


          


            

            

            

            

              Options
              
            


            

              What gets checked:
              Pivot counts & distributions across all topology and infrastructure columns. Formula integrity for A/Z labels, trunk labels, and hostname RIGHT() truncation. Port duplicate detection (hostname+port and location+port). Cabinet triplet naming convention (D-skip, length). Cable bundle/leg sequencing. Port numbering ranges per network/vendor. Colour and room consistency.
            


            
          

        


        
        

          

            

              
Activity

              
Recent validations

            

            
          

          

            
No validations yet.