Skip to content

README.md

Latest commit

 

History

History
118 lines (91 loc) · 6.18 KB

README.md

File metadata and controls

118 lines (91 loc) · 6.18 KB
Raw

Economic Research

Overview

The Economic Research use case produces structured economic research reports by coordinating multi-source data aggregation, indicator trend analysis, and professional research writing. It covers key macroeconomic indicators (GDP, inflation, employment, interest rates, trade balance), identifies correlations and leading/lagging relationships, generates forecasts with confidence levels, and produces actionable investment implications for capital markets analysts and portfolio managers.

Business Value

  • Multi-source aggregation -- automated collection and normalization of economic data across government agencies and financial databases
  • Trend detection -- identifies correlations, leading/lagging indicator relationships, and inflection points across macroeconomic datasets
  • Forecast generation -- produces forecasts with confidence levels and explicit time horizons for investment planning
  • Investment-ready output -- structured research reports with actionable recommendations and clear investment implications
  • Flexible scope -- supports full research, data-only, trend-only, report-only, or indicator-focused modes

Architecture

graph TB
    Request["Client Request"] --> Runtime["AgentCore Runtime"]
    Runtime --> Orchestrator["Orchestrator"]
    Orchestrator --> Aggregator["Data Aggregator<br/><small>Multi-source data collection</small>"]
    Orchestrator --> Trend["Trend Analyst<br/><small>Indicator trends & forecasts</small>"]
    Orchestrator --> Writer["Research Writer<br/><small>Structured reports & implications</small>"]
    Aggregator --> Bedrock["Amazon Bedrock<br/>(Claude)"]
    Trend --> Bedrock
    Writer --> Bedrock
    Aggregator --> S3["S3 Sample Data"]
    Trend --> S3
    Writer --> S3
    Aggregator --> Synthesis["Result Synthesis"]
    Trend --> Synthesis
    Writer --> Synthesis
    Synthesis --> Response["Response"]
Loading

Directory Structure

use_cases/economic_research/
├── README.md
└── src/
    └── strands/
        ├── __init__.py
        ├── config.py          # EconomicResearchSettings
        ├── models.py          # Pydantic request/response models
        ├── orchestrator.py    # EconomicResearchOrchestrator + run_economic_research()
        └── agents/
            ├── __init__.py
            ├── data_aggregator.py
            ├── trend_analyst.py
            └── research_writer.py

Agentic Design

The orchestrator uses a parallel fan-out pattern with five modes. In full mode, all three agents execute concurrently via asyncio.gather. In indicator_focus mode, the data aggregator and trend analyst run in parallel (without research writer). Individual modes (data_aggregation, trend_analysis, report_generation) invoke a single agent. The orchestrator synthesizes results through a structured prompt that produces JSON with economic overview, primary indicator, trend direction, key findings, and recommendations.

Agents

Agent Role Data Used Output
Data Aggregator Aggregates economic data from multiple sources (BEA, BLS, Fed, Treasury); normalizes datasets across formats and time periods; identifies data quality issues Entity profile via s3_retriever_tool Structured data summaries, normalized datasets, data quality notes
Trend Analyst Identifies trends across key indicators (GDP, inflation, employment, interest rates, trade balance); detects correlations and leading/lagging relationships; generates forecasts; flags inflection points Entity profile via s3_retriever_tool Trend directions, correlations, forecasts with confidence levels, inflection points
Research Writer Generates structured research reports; synthesizes data and trend analysis into coherent narratives; produces actionable insights and investment implications Entity profile via s3_retriever_tool Formatted research report with investment implications for capital markets analysts

Data and Tools

  • Tool: s3_retriever_tool -- retrieves economic research profiles and indicator data from S3
  • S3 data prefix: samples/economic_research/
  • Model: Claude Sonnet (via Amazon Bedrock), temperature 0.1, max 8192 tokens
  • Config thresholds: trend_confidence_threshold=0.75, max_data_sources=50, report_max_length=10000

Request / Response

Request -- ResearchRequest:

Field Type Description
entity_id str Research identifier (e.g., ECON001)
research_type ResearchType full, data_aggregation, trend_analysis, report_generation, indicator_focus
additional_context str | None Optional context

Response -- ResearchResponse:

Field Type Description
entity_id str Research identifier
research_id str Unique research UUID
timestamp datetime Research timestamp
economic_overview EconomicOverview | None Primary indicator, trend direction, data sources, key findings, correlations, forecast horizon
recommendations list[str] Actionable recommendations
summary str Executive summary
raw_analysis dict Raw agent output

Quick Start

# Deploy to AgentCore
USE_CASE_ID=economic_research ./scripts/deploy/full/deploy_agentcore.sh

# Test the deployment
./scripts/use_cases/economic_research/test/test_agentcore.sh

Sample Data

Located at data/samples/economic_research/

Entity ID Topic Region Description
ECON001 US Economic Outlook Q2 2026 United States GDP growth 2.3%, inflation 3.1%, unemployment 4.2%, fed funds rate 4.75%, data from BEA/BLS/Fed/Treasury, prior findings on moderating GDP, sticky services inflation, and gradual labor market cooling

Related Documentation