WORKSHOP GUIDE - SESSION 1

Local RAG & Privacy

Using Intel AI Playground to securely analyze confidential documents.

Part 1: Key Concepts

The Privacy Problem

Cloud AI (ChatGPT) stores your data. For government, legal, or health documents, this is unacceptable. Local AI runs 100% on your laptop's NPU, ensuring zero data leakage.

What is RAG? (Retrieval Augmented Generation)

Standard AI hallucinates facts. RAG forces the AI to read your specific PDF first, then answer based only on that document. It's like an open-book exam.

Part 2: The Lab - "The 500Cr Tender"

The Mission

You are a Bureaucrat. You have received a massive, confidential "Smart City RFP" document. You need to summarize it and check for compliance risks in 10 minutes without uploading it to the cloud.

The Evidence

Download the dummy tender document:

Tender_Document_Secret.pdf

Part 3: The Prompt Library

Level 1: The Executive Summary

Goal: Get the big picture instantly.

Summarize this tender document. What is the primary objective of the project and what is the estimated budget?

Level 2: The Technical Deep Dive

Goal: Extract specific technical requirements.

List all the mandatory IoT sensor requirements mentioned in the Technical Specifications section. Bullet points please.

Level 3: The Compliance Trap (Risk Check)

Goal: Find the hidden clause.

Check the 'Data Localization' clause. Does the tender require all data to be stored within India? Quote the exact text.

Verification: The AI should cite the page number (e.g. [Page 12]). Click the number to verify the source text yourself!

Level 4: The Penalty Clause

Goal: Financial risk assessment.

What represent the penalties for delaying the project by more than 3 months? Is there a force majeure clause?