Skip to content
English
Contact us
  • There are no suggestions because the search field is empty.

Supported File Types for ORIS AI BoQ Engine

What the AI reads and how to make the most of your files

The ORIS AI BoQ Engine reads your existing Bill of Quantities files to extract materials, quantities and units directly into ORIS. Understanding how the AI reads your file type helps you structure your data for the most accurate extraction with the least manual correction.

Microsoft Excel (.xlsx, .xls)

PDF and GAEB Files


Microsoft Excel (.xlsx, .xls)

Excel is the most common format for BoQ files and the format the AI BoQ Engine is best equipped to handle. The AI reads the full text content across all sheets in the workbook, interpreting structure, material descriptions, quantities and units from the content it finds.

HOW THE AI READS EXCEL FILES

  • The AI reads all sheets in the workbook together as a single body of content. It does not read sheet tab names as structural input; hierarchy is built from the text content within the sheets themselves, such as header rows and section titles.
  • A clear, prominent header row at the top of each sheet is the most reliable way to signal a top-level group to the AI. This header should name the section or package that sheet covers.
  • Row structure within each sheet is analysed to identify lower levels of the hierarchy, material line items, quantities, quantity units and any descriptive text associated with them.
  • Row structure within each sheet is analysed to identify lower levels of the hierarchy, material line items, quantities, quantity units and any descriptive text associated with them.
  • Blank rows and non-material rows such as totals, subtotals and section headers are generally detected and excluded, though they can increase processing time on large files.
  • The AI can detect and merge related rows where a description spans multiple lines, combining them into a single material entry.

WHAT THE AI EXTRACTS

  • Material description and name
  • Quantity value and unit (e.g. m3, tonnes, m2, linear metres)
  • Distinction between material items (A1-A3) from construction activities (A5)
  • The grouping and hierarchy inferred from the file structure

 

Supported

  • A clear, named header row at the top of each sheet
  • Distinct, descriptive section headers within sheets
  • Consistent column structure across sheets
  • Explicit quantity values in dedicated columns
  • Consistent unit labelling throughout

Not Supported

  • Relying on sheet tab names alone to define groups
  • Quantities embedded in text descriptions
  • Merged cells spanning multiple categories
  • Totals and sub-totals mixed into material rows without distinction
  • Multiple unrelated sections within a single sheet without clear separators

Note on large files

Files with a high number of rows or sheets will take longer to process. For very large BoQ files, consider splitting by discipline or package before uploading to reduce processing time and improve the accuracy of the extraction.


PDF and GAEB Files

Support for PDF and GAEB files is available in the AI BoQ Engine. PDF files are read using text extraction, which works best on digitally produced documents rather than scanned copies. GAEB is a structured tender format used primarily in German-speaking markets and is parsed directly from its structured data.

Tip

Regardless of file type, the same principle applies: the clearer and more consistent the data in your source file, the more accurate the AI extraction will be. Material naming, unit labelling and structural consistency all directly affect the quality of the output.

Database Coverage and Matching

When the AI maps extracted materials to a carbon database, the quality of the match depends on two things: how clearly the material is described in the file, and whether that material is covered in the selected database.

  • The AI can only match to materials that exist in the selected carbon database. If a material is not covered, no confident match is possible regardless of how well the file is structured. This is a database coverage limitation, not a file quality issue.
  • Where several database entries could match a material, the AI selects one by default. Always verify the selected match is appropriate for your project. For example, the AI may select a warm mix asphalt where your project uses a hot mix; these carry different carbon values.
  • When no suitable match is found, the AI may assign a Custom material placeholder. In this case, you will need to source and enter the carbon footprint value manually, from an EPD or another reliable reference, and document where it came from in the material description.