Tool 02 · Scoping · STAR

Requirements
Extraction

Upload any RFP, SoW, or specification. A²AI extracts every requirement, classifies it as Functional or Non-Functional, assigns MoSCoW priority, maps to SAP module, and confidence-scores each one.

Algorithm
TF-IDF + Pattern Classification
0
Requirements
0
Functional
0
Non-Functional
How It Works

Upload your document. The extraction engine scans every sentence for requirement indicators (shall, must, is required, needs to), then runs three parallel classifiers: Requirement Type (FR/NFR), MoSCoW Priority, and SAP Module. Confidence is scored by keyword density and sentence structure certainty. Low-confidence items surface in the Consultant Triage box.

Live Demo

Extract Requirements

Upload an RFP or specification document. Results appear instantly after extraction.

Document Upload
Upload RFP, SoW or Requirements Document
PDF · DOCX · XLSX · TXT · CSV — drag & drop or click
document.pdf remove
Upload a document to begin
Settings
Architecture

Extraction Algorithm

How TF-IDF and pattern classification turns unstructured text into structured requirements

Sentence Segmentation
Document is split on sentence boundaries (.!? ). Sentences shorter than 30 or longer than 350 characters are discarded as unlikely to be requirements.
Requirement Signal Detection
Each sentence is tested for modal verbs and requirement keywords: shall, must, will be required, should, is required, needs to, mandatory. Matches score a confidence boost of 20–40 points.
Three Parallel Classifiers
FR/NFR: Non-functional keywords (performance, security, availability) trigger NFR classification. MoSCoW: Inferred from strength of modal verbs. Module: SAP keyword scoring across 40 module vocabularies.