Security Tender AI Workflow

I need a repeatable, end-to-end AI workflow that speeds up electronic security estimating. Every week I receive full tender packs: the scope drawings and specifications come through as large PDFs and the quotation/BOM arrives in Excel. The job is for you to design, build and document a process that:

• reads every page of the PDF set, identifies the electronic security items (access control, CCTV, intercom, intrusion, etc.) and lifts the relevant quantities and notes;
• cross-checks those findings against the Excel BOM so I can see mismatches immediately;
• pushes the confirmed counts straight back into an estimating sheet;
• and finally produces a proof-of-count markup PDF. The markups must use Bluebeam security symbols so the client can visually verify that every device has been counted.

The workflow can be scripted in the language or platform you feel is most efficient—Python with OpenAI functions, Google AutoML, Power Automate, or another approach you suggest—as long as a non-coder like me can rerun it on future tenders with minimal setup. Clear documentation and a short screen-capture walk-through are part of the hand-over.

Accuracy and repeatability are the key acceptance tests: on a sample project I will compare your output counts and Bluebeam-style markups against a manual take-off; we will tweak once to reach parity and then the project is complete.