Tool 06 · Scoping · STAR

RFP
Chat

Upload one or more RFP or SoW documents and chat naturally. A²AI indexes every paragraph, retrieves the most relevant chunks, and synthesises answers grounded in your documents.

Algorithm
RAG · TF-IDF Retrieval · Chunk Synthesis
0
Documents
0
Chunks Indexed
0
Q&As
RAG Architecture

Documents are split into 300-word overlapping chunks. Each chunk is indexed using TF-IDF (term frequency-inverse document frequency). On each query, the top-3 most relevant chunks are retrieved by cosine similarity, then synthesised into a grounded answer. Sources are always cited.

Live Demo

Upload & Chat

Upload your RFP documents, then ask any question. No document? Try the suggested questions below.

Document Library
Upload Documents
PDF · DOCX · TXT — multiple allowed
No documents loaded — demo mode active
Chat Interface
What are the key deliverables?
What is the project timeline?
What SAP modules are in scope?
What are the compliance requirements?
Hello! I am ready to answer questions about your RFP documents. Upload documents above or use the suggested questions to see a demo response.
A²AI RFP Assistant — RAG-powered
Architecture

RAG Pipeline

How documents become a queryable knowledge base

Document Chunking
Each document is split into 300-word overlapping chunks (50-word overlap) to preserve context across boundaries. Chunks are labelled with source filename and page estimate.
TF-IDF Retrieval
Query terms are scored against every chunk using TF-IDF cosine similarity. Top-3 chunks above a relevance threshold are retrieved. Stop-words are removed before scoring.
Grounded Synthesis
The retrieved chunks are synthesised into a structured answer. Every factual claim is anchored to a source chunk, preventing hallucination on out-of-corpus questions.