Estimating software should handle the file types estimators actually receive: PDF drawings, CAD, images, spreadsheets, DOCX/XLSX, text, ZIP bundles, and email files. Unsupported files should stay visible instead of disappearing.
What file types
Estimating software should support the RFQ file types estimators actually receive: PDF drawings, DXF and DWG CAD exports, STEP/STP and IGES/IGS models, STL/OBJ/3MF/IFC files, images, spreadsheets, DOCX documents, text and data files, ZIP bundles, and email files. Just as important, unsupported or unreadable files must stay visible as issues so scope is not missed during pricing.
The best file-handling workflow does three things: opens common formats in the project workspace, preserves source evidence for every extracted item, and clearly flags anything skipped, unreadable, encrypted, corrupted, or externally reviewed. That protects the estimator from assuming the file set was fully processed when it was not.
For the broader intake workflow around file completeness, see the RFQ intake checklist for fabrication estimating. For file organisation before estimating, see how to organise RFQ files before estimating.
File types estimators actually
Estimators in fabrication and construction-adjacent work handle a mixed file set on every RFQ. PDF drawings are the most common format for scope definition and dimensional references. These prints serve as the primary contract document because why technical drawings are your legal contract remains the baseline for resolving component tolerances and manufacturing revisions.
Images are used for markups, site photos, reference conditions, and quality notes. Spreadsheets hold bills of quantities, material schedules, and commercial summary sheets. Text specifications, DOCX documents, XML/JSON data files, and email files can contain scope descriptions, coating schedules, addenda, and commercial terms.
For defence fabricators, local CAD rendering is a compliance requirement. Cloud SaaS estimators require files to be uploaded to external platforms, which introduces data sovereignty issues. Local-first desktop rendering keeps all CAD files, drawings, and IP within the secure local network, maintaining CMMC and ITAR compliance.
Happens when file types
Not every RFQ file fits neatly into supported formats. Proprietary CAD formats, RAR archives, legacy DOC/XLS files, presentation decks, scanned handwritten notes, encrypted documents, or corrupt attachments may be unreadable by the estimating software. The key requirement is that unsupported or unreadable files are flagged explicitly rather than skipped silently.
If a file is not processed but contains scope information, the estimator needs to know. A skipped file marker in the project workspace tells the estimator to review that file separately before the quote goes out. Unsupported file types that contain pricing or scope data create hidden risks if they are not accounted for.
For each unsupported file, the estimator should have the option to add a manual note, flag it for clarification, or attach it to the project as a reference that was reviewed externally. The project file register should track which files were processed, which were skipped, and why.
Supported file formats fabrication
The specific support level for each format depends on the platform. The core principle is that unsupported or unreadable files are surfaced as issues rather than disappearing into the project folder.
AI preprocessing handles different
AI preprocessing can use readable PDFs, text files, DOCX/XLSX content, email bodies, and prepared image evidence to draft a first-pass file summary or clarification list. For images, the preparation step matters: clean screenshots and markups are usually more useful than blurred site photos or low-resolution scans.
For CAD-derived formats, AI preprocessing is more limited. The current product contract treats CAD as render or metadata evidence when the file can be opened. That is useful for review context, but it is not the same as a human estimator checking fabrication intent, weld symbols, coating notes, tolerances, or exclusions.
The same boundary applies to all formats: AI assists with extraction and grouping, but the estimator reviews every line item against the source document before it reaches the customer quote. For the review model, see why AI estimating should assist, not replace estimators and what local-first estimating software changes for sensitive RFQ data.
File support
File type support tends to drift when it is documented in multiple places. The frontend may show a file picker with one filter, the service layer may handle a different set, and the documentation may describe a third. For estimating teams, this creates confusion about what files can be processed and what will be skipped.
The solution is a single file support contract that defines every supported and unsupported format, how each is handled, and what happens when an unsupported file is encountered. In the current Kwantflow app, that contract drives upload and attach rules, preview type, thumbnail priority, MIME inference, and Estimate AI eligibility from the same source.
That matters operationally. If DOCX is allowed but legacy DOC is not, or ZIP is accepted but RAR is not, the estimator should see that difference before pricing begins. For a practical RFQ file handling workflow that covers intake through supported file review, see the RFQ intake checklist for fabrication estimating.
Unreadable file escalation workflow
When a file is unreadable, the workflow should not stop silently. Mark the file as unsupported, unreadable, encrypted, corrupted, or externally reviewed. Then decide whether it is scope-critical. A corrupted site photo might be low impact. A locked specification or missing CAD assembly can be quote-critical and should trigger clarification before pricing proceeds.
Examples by trade RFQ
Structural steel RFQ: the estimator may receive GA drawings as PDFs, connection details as PDFs, a member schedule in XLSX, and model exports in STEP or IFC. The file-handling workflow should keep the drawing register, member schedule, and model references linked so a changed drawing can be traced to affected quantities.
Sheet metal RFQ: the estimator may receive DXF files for cutting, PDFs for folded part drawings, photos for finish references, and emails describing hardware or packaging. Using limited dimension drawings for quoting CNC jobs allows the estimator to extract folded geometries rapidly without full drafting sets.
Subcontract package: the file set may include specifications, addenda, supplier quotes, marked-up PDFs, email attachments, and compressed ZIP bundles. Unsupported archives, encrypted specs, or legacy office files should be visible issues because they can contain scope, compliance, or commercial terms that affect the quote.
Buyer checklist
When evaluating estimating software, ask vendors to process a real RFQ pack rather than a clean demo folder. Include a mixed set of PDFs, CAD, spreadsheets, images, office documents, poor scans, duplicate files, and unsupported formats. The goal is to see what the tool does when the file set is messy, because real RFQs are messy.
Strong file handling is a prerequisite for RFQ processing software before pricing. If the system cannot clearly show which files were processed and which were skipped, it is not ready to support automated intake or AI-assisted extraction.
Ways estimators can keep quote review clear:
- Estimating software should support the RFQ formats estimators actually receive: PDF, DXF, DWG, STEP/STP, IGES/IGS, STL, OBJ, 3MF, IFC, images, DOCX/XLSX, text/data files, ZIP bundles, and email files.
- Unsupported or unreadable files should be marked as issues, not ignored. Legacy DOC/XLS files, PPTX decks, RAR archives, encrypted files, and corrupt files can all hide scope if they disappear from the register.
- AI preprocessing can use PDFs, images, text, DOCX/XLSX, email bodies, and CAD renders where the file is readable, but the estimate should still show which source file each line item came from.
- File handling contracts should be documented in one place so upload, preview, thumbnail, and AI eligibility behaviour stay consistent.

