Financial Scan - ReAct Agent for SEC Filing Analysis

Technology Stack

Agent Framework LangChain ReAct

LLM OpenAI GPT-4

Vector DB ChromaDB

Data Source SEC EDGAR API

Backend FastAPI + Uvicorn

Architecture Note: ReAct (Reasoning + Acting) agent pattern with RAG over SEC filings for financial research and company analysis.

RAG Pipeline Components

📄

Document Loader

SEC Filing Ingestion

✂️

Text Splitter

Recursive Chunking

🔢

Embeddings

OpenAI text-embedding-ada

🗄️

Vector Store

ChromaDB Persistence

🔍

Retriever

Similarity Search

🤖

ReAct Agent

Reasoning + Acting

SEC Filing Ingestion Pipeline

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#C17852', 'primaryTextColor': '#F0F6FC', 'primaryBorderColor': '#4A5E32', 'lineColor': '#E6C98F', 'secondaryColor': '#161B22', 'tertiaryColor': '#0D1117', 'background': '#0D1117', 'mainBkg': '#161B22', 'nodeBorder': '#4A5E32', 'clusterBkg': '#161B22', 'clusterBorder': '#4A5E32', 'titleColor': '#E6C98F', 'edgeLabelBackground': '#161B22'}}}%%
flowchart LR
    subgraph Input["📥 Data Sources"]
        T[("📊 Stock Ticker")]
        F[("📄 Filing Type
10-K, 10-Q, 8-K")]
    end

    subgraph Edgar["🏛️ EDGAR API"]
        A["🔍 Search Filings"]
        D["📥 Download HTML"]
        P["🔧 Parse Content"]
    end

    subgraph Process["⚙️ Processing"]
        C["✂️ Recursive Splitter
chunk_size=1000"]
        E["🔢 OpenAI Embeddings
text-embedding-ada"]
    end

    subgraph Storage["💾 Vector Store"]
        V[("🗄️ ChromaDB
Persistent")]
        M[("📋 Metadata
ticker, date, type")]
    end

    T --> A
    F --> A
    A --> D
    D --> P
    P --> C
    C --> E
    E --> V
    E --> M

RAG Query Pipeline

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#C17852', 'primaryTextColor': '#F0F6FC', 'primaryBorderColor': '#4A5E32', 'lineColor': '#E6C98F', 'secondaryColor': '#161B22', 'tertiaryColor': '#0D1117', 'background': '#0D1117', 'mainBkg': '#161B22', 'nodeBorder': '#4A5E32', 'clusterBkg': '#161B22', 'clusterBorder': '#4A5E32', 'titleColor': '#E6C98F', 'edgeLabelBackground': '#161B22'}}}%%
flowchart TB
    subgraph Query["💬 User Query"]
        Q["🔍 Financial Question"]
    end

    subgraph Embed["🔢 Query Processing"]
        QE["🔢 Query Embedding"]
    end

    subgraph Retrieval["📚 Retrieval"]
        VS[("🗄️ ChromaDB")]
        SR["🎯 Similarity Search
top_k=5"]
        RD["📄 Retrieved Docs"]
    end

    subgraph Context["📋 Context Building"]
        CB["📝 Context Builder"]
        MP["📊 Metadata Filter"]
    end

    subgraph LLM["🧠 Generation"]
        PT["📝 Prompt Template"]
        G["🤖 GPT-4
Generate Answer"]
    end

    subgraph Output["📤 Response"]
        A["✅ Answer + Citations"]
    end

    Q --> QE
    QE --> VS
    VS --> SR
    SR --> RD
    RD --> CB
    MP --> CB
    CB --> PT
    PT --> G
    G --> A

ReAct Agent Loop

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#C17852', 'primaryTextColor': '#F0F6FC', 'primaryBorderColor': '#4A5E32', 'lineColor': '#E6C98F', 'secondaryColor': '#161B22', 'tertiaryColor': '#0D1117', 'background': '#0D1117', 'mainBkg': '#161B22', 'nodeBorder': '#4A5E32', 'clusterBkg': '#161B22', 'clusterBorder': '#4A5E32', 'titleColor': '#E6C98F', 'edgeLabelBackground': '#161B22'}}}%%
flowchart TB
    subgraph Input["📥 Input"]
        Q["🔍 Complex Query"]
    end

    subgraph Agent["🤖 ReAct Agent"]
        T["💭 Thought
Reasoning Step"]
        A["⚡ Action
Choose Tool"]
        O["👁️ Observation
Tool Result"]
        D{"🎯 Done?"}
    end

    subgraph Tools["🔧 Available Tools"]
        R["📚 RAG Search
Vector Retrieval"]
        E["🏛️ EDGAR Query
Filing Search"]
        C["🔢 Calculator
Financial Math"]
    end

    subgraph Output["📤 Output"]
        F["✅ Final Answer"]
    end

    Q --> T
    T --> A
    A --> R
    A --> E
    A --> C
    R --> O
    E --> O
    C --> O
    O --> D
    D -->|"No"| T
    D -->|"Yes"| F

Financial Metrics Extraction

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#C17852', 'primaryTextColor': '#F0F6FC', 'primaryBorderColor': '#4A5E32', 'lineColor': '#E6C98F', 'secondaryColor': '#161B22', 'tertiaryColor': '#0D1117', 'background': '#0D1117', 'mainBkg': '#161B22', 'nodeBorder': '#4A5E32', 'clusterBkg': '#161B22', 'clusterBorder': '#4A5E32', 'titleColor': '#E6C98F', 'edgeLabelBackground': '#161B22'}}}%%
flowchart LR
    subgraph Filing["📄 SEC Filing"]
        F10["📊 10-K Annual"]
        F10Q["📊 10-Q Quarterly"]
    end

    subgraph Extract["⚙️ Extraction"]
        IS["💰 Income Statement"]
        BS["📋 Balance Sheet"]
        CF["💸 Cash Flow"]
    end

    subgraph Metrics["📈 Calculated Metrics"]
        REV["💵 Revenue Growth"]
        NI["📊 Net Income"]
        EPS["📈 EPS"]
        ROE["📉 ROE / ROA"]
        DR["📊 Debt Ratio"]
    end

    subgraph Store["💾 Storage"]
        VDB[("🗄️ ChromaDB")]
        JSON[("📝 Structured JSON")]
    end

    F10 --> IS
    F10 --> BS
    F10 --> CF
    F10Q --> IS
    F10Q --> BS
    IS --> REV
    IS --> NI
    IS --> EPS
    BS --> ROE
    BS --> DR
    CF --> DR
    REV --> VDB
    NI --> VDB
    EPS --> JSON
    ROE --> JSON
    DR --> JSON

Multi-Company Analysis

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#C17852', 'primaryTextColor': '#F0F6FC', 'primaryBorderColor': '#4A5E32', 'lineColor': '#E6C98F', 'secondaryColor': '#161B22', 'tertiaryColor': '#0D1117', 'background': '#0D1117', 'mainBkg': '#161B22', 'nodeBorder': '#4A5E32', 'clusterBkg': '#161B22', 'clusterBorder': '#4A5E32', 'titleColor': '#E6C98F', 'edgeLabelBackground': '#161B22'}}}%%
flowchart TB
    subgraph Input["📥 Input"]
        T1["📊 Ticker A"]
        T2["📊 Ticker B"]
        Q["🔍 Comparison Query"]
    end

    subgraph Parallel["⚡ Parallel Retrieval"]
        direction TB
        R1["📚 RAG: Company A"]
        R2["📚 RAG: Company B"]
    end

    subgraph Context["📋 Context Merge"]
        M["🔗 Merge Documents"]
        D["📊 Extract Metrics"]
    end

    subgraph Analysis["🧠 GPT-4 Analysis"]
        P["📝 Comparison Prompt"]
        A["🤖 Generate Analysis"]
    end

    subgraph Output["📤 Output"]
        C["📊 Comparison Report"]
        R["📈 Relative Metrics"]
    end

    T1 --> R1
    T2 --> R2
    Q --> P
    R1 --> M
    R2 --> M
    M --> D
    D --> P
    P --> A
    A --> C
    A --> R

Complete System Architecture

%%{init: {'theme': 'dark', 'themeVariables': { 'primaryColor': '#C17852', 'primaryTextColor': '#F0F6FC', 'primaryBorderColor': '#4A5E32', 'lineColor': '#E6C98F', 'secondaryColor': '#161B22', 'tertiaryColor': '#0D1117', 'background': '#0D1117', 'mainBkg': '#161B22', 'nodeBorder': '#4A5E32', 'clusterBkg': '#161B22', 'clusterBorder': '#4A5E32', 'titleColor': '#E6C98F', 'edgeLabelBackground': '#161B22'}}}%%
flowchart TB
    subgraph Client["🖥️ Frontend"]
        UI["💻 Web Interface"]
    end

    subgraph API["🔌 FastAPI Backend"]
        EP["🛤️ Endpoints"]
        AUTH["🔐 Auth"]
    end

    subgraph Agent["🤖 LangChain Agent"]
        RA["🧠 ReAct Agent"]
        TK["🔧 Tool Registry"]
    end

    subgraph RAG["📚 RAG Layer"]
        RT["🔍 Retriever"]
        EM["🔢 Embeddings"]
    end

    subgraph External["🌐 External APIs"]
        ED["🏛️ SEC EDGAR"]
        OAI["🤖 OpenAI API"]
    end

    subgraph Storage["💾 Persistence"]
        CH[("🗄️ ChromaDB")]
        FS[("📁 File System")]
    end

    UI <--> EP
    EP --> AUTH
    AUTH --> RA
    RA <--> TK
    TK <--> RT
    RT <--> EM
    EM <--> OAI
    RT <--> CH
    TK <--> ED
    ED --> FS