Multi-Channel Service AI System Architecture Design

prj_mcs Project

prj_mcs is an AI system prototype for Multi-Channel Service. The initial design was developed over April–May 2025 across 2 commits.

Multi-Channel Architecture

graph TD
    A[API Gateway] --> B[채널 라우터]
    B --> C[웹 채널]
    B --> D[모바일 채널]
    B --> E[챗봇 채널]

    C --> F[AI 엔진]
    D --> F
    E --> F

    F --> G[응답 포맷터]
    G --> C
    G --> D
    G --> E

Per-Channel Response Format

class ResponseFormatter:
    def format(self, response: str, channel: str):
        if channel == "web":
            return {"html": markdown_to_html(response)}
        elif channel == "mobile":
            return {"text": response, "cards": extract_cards(response)}
        elif channel == "chatbot":
            return {"message": truncate(response, 500)}

Core Design Principles

1. Channel independence: Business logic is decoupled from the channel layer
2. Response adaptation: Automatic format conversion tailored to each channel
3. Shared AI engine: All channels use the same underlying AI engine

It was a small, 2-commit project, but it was a good opportunity to nail down design patterns for multi-channel service delivery.