AI Technical Bid Comparator

Заказчик: AI | Опубликовано: 04.02.2026

I have a repeatable workflow that starts with my own technical specification (today the example is a simple pen, tomorrow it will be pumps, valves or full-scale equipment). I upload that spec, then I receive three vendor proposals in their own templates—some come as PDFs, others as Word documents or Excel sheets. The tool you build must accept all three formats, read each proposal, and automatically generate a single comparative Excel sheet that lines items side-by-side so I can see who matches the requirement and where they differ. Because vendors rarely use identical wording, I need contextual AI/NLP built in. If one bidder says “cap” and another writes “nib cover,” the system should recognise they refer to the same component, align them in the comparison, and flag any real deviations. Context understanding is the critical capability; synonym matching and term normalisation are welcome if they improve accuracy, but the core is to make the comparison intelligent rather than literal. Workflow I envision 1. Upload my technical spec (PDF/Word/Excel). 2. Upload each vendor file (again, PDF, Word, or Excel). 3. System parses, maps terms, and asks me to confirm any uncertain matches. 4. It outputs a clean, well-formatted Excel workbook showing requirement vs. each vendor’s offer, along with an automated “best-fit” recommendation. Future iterations may add scoring weightings, but the first milestone is the Excel comparison itself. Tech is up to you—Python with pandas, open-source NLP libraries such as spaCy or transformers, or a lightweight web front end that calls a backend service. Whatever stack you choose, the deliverable must run on a standard Windows machine or be easily deployed to a small cloud instance. Acceptance criteria • Uploads accepted: PDF, Word, Excel. • Comparative sheet created in .xlsx with no manual cleanup. • AI mapping reaches ≥90 % correct term alignment in a sample set I provide. • Clear documentation and basic UI/CLI instructions so I can repeat the process for new products. If you’ve tackled similar document harmonisation or procurement tools, let me see a quick demo or code snippet, and we can move straight into the build.