Requirements Analyzer

This tool leverages a Hybrid AI Architecture to transform unstructured text into actionable project artifacts. By combining deep semantic understanding with linguistic precision, it identifies actors, requirements, and constraints with high accuracy.

• Semantic Intelligence (Sentence Transformers): Powered by the multi-qa-MiniLM-L6-cos-v1 model, the system understands intent rather than just keywords. It can identify a "Requirement" even if the word itself never appears.
• Linguistic Precision (spaCy): The en_core_web_sm NLP engine performs deep dependency parsing to identify "Agents" (who is acting) and "Objects" (what is being acted upon), ensuring extracted actors are grammatically relevant.
• Neural Semantic Search: The tool converts your text into high-dimensional vectors. It then uses Cosine Similarity to match your sentences against AI "Probes"—questions designed to find specific types of information.
• Logical Reasoning (Cross-Encoders): Powered by nli-distilroberta-base, the AI performs Natural Language Inference. Unlike basic search, it can detect if two statements logically support or contradict one another, helping you catch hidden project conflicts.
• Zero-Shot Extraction: Unlike traditional models, this AI does not need to be "trained" on your specific data. It uses its pre-trained knowledge to find artifacts in any domain immediately.

This approach prioritises Explainability and Control. You can view and modify the AI's "Probes" in the settings, giving you direct influence over how the neural network interprets your documentation.

info Quick Start Guide

1. Paste or Edit: Use the TinyMCE editor to paste project notes or documentation. The AI will handle the HTML formatting automatically.
2. Select Artifacts: Choose the categories you wish to extract (e.g., Requirements, Scenarios).
3. Tune Probes: (Optional) Use the Advanced Settings to modify the AI's search prompts if you are looking for highly specific or niche information.
4. Analyse: The model will process the text sentence-by-sentence to ensure no details are missed.
5. Refine & Export: Review the results in the scrollable tables, edit, check duplicate, check conflicts, remove any "noise," and export your structured data to CSV for Jira or Excel.