Update Readme for REPORT

[anshk8]

MUST HAVE

• Report.md: You must include a report with Reasoning, design, what you did, anything relevant for reviewer
• requirements.txt - All versions etc
• runtime.txt - Python Version

AC:

Detailed Report

My Architecture: Details what my agents do, what techniques each use, example workflow (I can take an example from tests). This will break down how my Agent is Accurate and Also Robust (See Eval Criteria on Repo)

**How to run it: **

Code Organization: Mention how I have helpers, prompts, pydantic schemas etc

Challenges Faced: Mention about how I did Multi-Agent, research finds how multi agent is not the best for this, error carries over. Mention about size of data and how it is small. Link to the Branch with MultiAgent

Major Learning: Mention how my solution used advanced Techniques, cOt, FewShotExamples, IF I have time I will include a multi conversational Feature and/or generate k queries. MAKE SURE TO SHOW IMAGE OF GITHUB CONTRIBUTION GRAPH and how I learned to stay consistent coding everyday

About me: A little section at the end about me

Test Execution

• Create a new Venv, run pip install -r requirements.txt and make sure the main.py works perfectly

[anshk8]

Why my Solution Stands out:
- Clean Code Organization with Pydantic Schemas, Proper Helpers, Prompts, Graph Folder, Agents Organized for Clear Cod
- Modern Agentic Framework using Langgraph, making use of State and Conditional Routes instead of a pile of If statements
- Innovative Features, KCandidateQueries with Temperature Adjustments and Multi-Conversational Support for Ambiguous Queries

• Special Techniqes cOt Prompts, FewShotExamples, Correction Loops, Proper Context
• Trial and Error With Different Workflows: Went from a 3-Agent Approach to more accurate Smaller.

[anshk8]

Add Next Steps:

• Security (Prevent SQL Injection checking Multiconversational)
• More Few Shot Examples Bigger Bank