InsightBestAI

🚀 Demo & Live Application

📽️ 1-Minute Demo (Recommended)

Click Here to Watch the Full Project Demo on YouTube

This video shows the complete functionality, as the live app may be slow due to API rate limits.

🔗 Live Application

You can also try the live application here: Try InsightBestAI on Streamlit

⚠️ Important Note: This app runs on a free Streamlit tier and connects to the live Gemini API. Due to shared IP rate limits on the free tier, the app may be very slow or take a long time to load. For a guaranteed, 1-minute overview, please watch the demo video above!

---

📌 Overview

InsightBestAI is an AI-powered tool that streamlines financial risk analysis for professionals. Instead of manually reading through hundreds of pages of reports, articles, blogs, and PDFs, this platform scans and summarizes all relevant sources on the internet using advanced LLMs.

Whether you're an investor, analyst, or ESG-focused stakeholder, InsightBestAI helps you focus on decisions—not data gathering.

---

🚀 Features

• 🔍 Financial Risk Analysis (Credit Risk, Operational Risk, Compliance Risk, Strategic Risk)
• ♻️ ESG Insights (Environmental, Social, Governance)
• 🌐 Force-Directed Network Graph showing risk interdependencies
• 🧠 Powered by Google's Gemini 2.5 Flash with grounded search

---

🧱 Tech Stack

Backend

• FastAPI
• LangGraph
• LangChain
• Gemini 2.5 Flash (via Gemini API)
• Pydantic

Frontend

• Streamlit
• streamlit-searchbox

External Services

• Google Gemini API with grounded web search

---

🧠 Multi-Agent Architecture Overview

The architecture behind InsightBestAI is designed as a modular multi-agent system built on top of LangGraph, Streamlit, and Google Gemini 2.5. It executes three intelligent agents in a coordinated flow, enabling robust, explainable, and structured analysis of financial and ESG risks, along with knowledge graph generation.

---

🔄 Architecture Diagram

---

🎯 Input

The user provides a listed company name through a simple Streamlit UI.

---

🛠️ Agent Responsibilities & Workflow

1. 🔍 Financial Risk Assessment Agent (risk_reporter.py)

Purpose: Evaluates the financial health and vulnerabilities of a company.

Categories Analyzed:

• Credit Risk
• Operational Risk
• Strategic Risk
• Compliance Risk

Powered By: Gemini 2.5 Flash + ReAct agent + grounded web search

Output: For each risk category, the agent generates:

• risk_title
• description
• severity (High, Medium, Low)
• impact and mitigation
• Citations with URLs

🧩 This is the foundational agent whose structured risk outputs are used later for graph generation.

---

2. ♻️ ESG Reporting Agent (esg_reporting.py)

Purpose: Produces ESG (Environmental, Social, Governance) insights.

Execution: Runs three parallel threads for each ESG dimension.

Powered By: Gemini 2.5 Flash + ReAct agent + grounded web search

Output Per Category:

• ESG category
• Rich narrative ESG description
• List of source citations

📚 Helps users understand a company's sustainability, ethics, and governance profile using reliable, cited data.

---

3. 🌐 Knowledge Graph Generator (knowledge_graph.py)

Purpose: Builds a force-directed dependency graph showing logical interconnections among the financial risks.

Input: Uses the textual descriptions produced by the Financial Risk Agent.

Powered By: Gemini 2.5 Pro with structured output parsing

Output Schema:

• nodes: Each node represents a unique risk (with id, name, description)
• links: Directed edges showing how risks influence or depend on each other

🔗 Enables users to visualize and understand how risks are semantically related.

---

🔄 Execution Flow

flowchart TD
    A[🏢 Company Name Input] --> B[🔍 Financial Risk Agent]
    A --> C[♻️ ESG Risk Agent]
    B --> D[🌐 Knowledge Graph Agent]
    B --> E[📊 Risk Table Output]
    C --> F[📘 ESG Report Output]
    D --> G[📡 Force-Directed Graph Output]

Loading

---

⚙️ Getting Started

1. Install Dependencies

Make sure you have Python installed. Then run:

uv sync  

2. Start Backend (FastAPI)

uv run uvicorn agents.main:app --reload

3. Start Frontend (Streamlit UI)

uv run streamlit run ui.py

---

🔧 Usage

Input

• Type a listed company name into the search bar

Output

• 🔹 Financial Risks
• 🔹 ESG Metrics & Insights
• 🔹 Force-Directed Graph of risk category dependencies

---

⚠️ Known Limitations

• ⏳ Output generation may take time due to Gemini's strict rate limits
• 🧾 Markdown rendering inconsistencies due to Streamlit's internal markdown parsing in force-directed graph explanation box

---

🧭 Future Improvements

• ⚡ Improve performance with Gemini Pro (paid tier)
• 🔗 Add support for Gemini URL-context for more grounded and document-specific analysis
• 🔝 Upgrading to the paid tier also helps overcome rate limit restrictions and results in more accurate, grounded outputs.

---