What is Amazon Bedrock Agents Financial Services Examples

amazon-bedrock-agents-financial-services-examples is a collection of examples demonstrating the use of Inline Agents and Model Context Protocol (MCP) servers specifically tailored for the Financial Services Industry (FSI) using Amazon Bedrock Agents.

Use cases

Use cases include automating financial data analysis, developing insurance underwriting models, and creating trading algorithms for capital markets, leveraging the capabilities of Amazon Bedrock Agents.

How to use

To use amazon-bedrock-agents-financial-services-examples, first install the required dependencies by navigating to the InlineAgent directory and running ‘pip install -e .’. Then, explore the main.py files in the mcp_servers directories to understand the functionality of each MCP server before diving into the industry-specific examples.

Key features

Key features include a variety of MCP servers such as a Python REPL for code execution, access to financial datasets, and examples tailored for insurance and capital markets, all designed to facilitate the integration of Amazon Bedrock Agents in financial services.

Where to use

amazon-bedrock-agents-financial-services-examples can be utilized in various sectors within the Financial Services Industry, including banking, insurance, and capital markets, where automation and data analysis are crucial.

🤖 Amazon Bedrock Agents FSI Examples with MCP

📋 Table of Contents

• Installation
• Getting Started
• Repository Structure
  • MCP Servers
  • Managed Tools
  • Insurance Examples
  • Capital Markets Examples

🔧 Installation

1. Ensure you have Python 3.11 or higher installed in your system.

# Check your Python version.
python --version

2. Install uv, a Python package and project manager, if not already installed. Ensure that they are added to your PATH.

# On macOS and Linux.
curl -LsSf https://astral.sh/uv/install.sh | sh

# On Windows.
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

You can confirm the installation with uv --version afterwards.

3. Create a virtual environment for project.

uv venv

This will create a virutal environment called .venv in the project’s root directory.

4. Activate the virtual environment.

# On macOS and Linux.
source .venv/bin/activate

# On Windows.
.venv\Scripts\activate

5. Install the required packages.

cd src/InlineAgent
uv pip install -e .

6. Set up each Model Context Protocol (MCP) server under mcp_servers directory.

Explore the README.md for each MCP server to understand how to set up the MCP server. Additionally, you can review the main.py to understand each MCP server and how it can be used.

# Review the README.md to understand the project architecture.
cd mcp_servers

# Review the README.md and main.py to set up the MCP server and learn how it works.
cd mcp_servers/python-repl

# Repeat for other MCP servers.

💡 Understanding these individual MCPs will make it easier to comprehend the more complex industry-specific examples, which typically use multiple MCPs together.

🏗️ Repository Structure

🔌 MCP Servers

The mcp_servers directory contains various Model Context Protocol (MCP) servers:

• 🐍 python-repl: Python REPL environment for executing code in a persistent environment
• 📊 financial-datasets: Access to financial market data for stocks, cryptocurrencies, and other instruments
• 🔍 bedrock-kb-search: Amazon Bedrock Knowledge Bases search functionality
• 📈 fredapi: Federal Reserve Economic Data API integration for economic indicators
• 🔎 perplexity-search: Perplexity Search service for web information retrieval
• 📁 filesystem: File system operations for reading and writing files

🛠️ Managed Tools

The managed_tools directory includes:

• 👨‍💻 code_interpreter: The Managed Code Interpreter for Amazon Bedrock Agents for executing Python code and data visualization

🏥 Insurance

The insurance directory contains insurance industry specific examples:

📋 actuarial_modelling_assistant

What it does: Analyzes insurance datasets to identify trends, model risks, and generate actuarial insights.

MCPs used:

• 🐍 Python REPL: For data analysis, statistical modeling, and visualization
• 📁 Filesystem: Mentioned in the README for file operations, though not explicitly configured in config.py

Key features:

• Exploratory data analysis on policy, claims, and risk data
• Statistical modeling for claim frequency and severity
• Loss ratio and reserve adequacy calculations
• Actuarial visualizations and reports

📝 rate_filing_comparison

What it does: Compares insurance rate filings from different insurers to identify differences and market trends.

MCPs used:

• 🔍 Bedrock KB Search: For retrieving rate filing documents from knowledge bases
• 📁 Filesystem: For saving comparison reports to output directories

Key features:

• Knowledge base search of insurance rate filings
• Detailed comparison of premium changes
• Coverage modification analysis
• Markdown report generation

💹 Capital Markets

The capital_markets directory includes capital markets related examples:

💰 crypto_investment_agent

What it does: Analyzes cryptocurrency investment opportunities and provides investment recommendations.

MCPs used:

• 📊 Financial Datasets: For cryptocurrency price data
• 📈 FRED API: For macroeconomic indicators
• 🔎 Perplexity Search: For market news and sentiment

Managed tools used:

• 👨‍💻 Code Interpreter: For investment modeling, risk analysis, and data visualization

Key features:

• Historical cryptocurrency price analysis
• Macroeconomic impact assessment
• Risk modeling and scenario simulation
• Investment allocation recommendations

📉 stock_data_processing

What it does: Processes stock market data to identify technical patterns and develop trading strategies.

MCPs used:

• 📊 Financial Datasets: For stock market data
• 📁 Filesystem: For storing results and trading signals

Managed tools used:

• 👨‍💻 Code Interpreter: For technical analysis and strategy backtesting

Key features:

• Technical indicator calculation
• Trading pattern identification
• Strategy backtesting and optimization
• Performance visualization

📊 historical_macro

What it does: Identifies historical periods with macroeconomic conditions similar to the present.

MCPs used:

• 📈 FRED API: For economic data retrieval
• 🔎 Perplexity Search: For historical context and market research

Key features:

• Economic indicator comparison
• Historical parallel identification
• Similarity scoring and ranking
• Forward-looking insights based on historical patterns