Blog

In this blog, we introduces the latest technologies and research related to SimpleModeling, as well as ongoing initiatives and the process of trial and error in development. Through these articles, we share the practical possibilities of Literate Model-Driven Development and AI-Driven Development.

img-blur-shadow-blog-2
Blog

My Personal AI-Driven Development

Rather than using generative AI merely as a code generation tool, this article introduces a development style in which design documents for specifications, source code, and executable specifications are evolved together as an integrated whole. The central themes of this article are the design of constraints that prevent AI from acting too freely, the clear separation between responsibilities best handled by humans and those delegated to AI, and documentation strategies for accumulating and reusing knowledge. Read More

2026-01-05

img-blur-shadow-blog-2
Blog

Semantic Integration Engine CLI

The Semantic Integration Engine is a multifaceted semantic search engine that provides multiple interfaces, including REST, MCP, and ChatGPT WebSocket. Among these, the REST API is positioned as the most general-purpose and stable interface for external exposure. In this article, we demonstrated how to build a CLI on top of the REST API, enabling SIE’s query functionality to be used directly from the console and shell scripts. We also confirmed the potential of combining this CLI with VS Code Codex and generative AI, allowing it to be used as a tool for acquiring external knowledge. Read More

2025-12-29

img-blur-shadow
Blog Integration between the Semantic Integration Engine and VSCode

The Semantic Integration Engine is designed as a RAG platform that integrates RDF, vector search, and knowledge graphs. In this article, with the goal of using SIE from VSCode via MCP, we present the assumed MCP specifications, architecture, a demo environment built with docker compose, and concrete examples of JSON-RPC–based MCP communication. At present, because MCP registration on the VSCode side did not work as expected, verification is performed using a pseudo session; nevertheless, the article provides a concrete understanding of how SIE behaves as an MCP server for VSCode.

2025-12-22

img-blur-shadow
Blog Integration between the Semantic Integration Engine and ChatGPT

This article explains, at the protocol level, what happens when the Semantic Integration Engine and ChatGPT are integrated via MCP (Model Context Protocol), focusing on how ChatGPT uses SIE’s knowledge, performs reasoning, and generates the final response. It provides a detailed explanation from a protocol-level perspective. The important point is that ChatGPT is not requesting “the answer itself” from SIE, but rather retrieving the materials and evidence it needs to reason on its own. In this article, we recreate a simulated MCP session, and examine why SIE’s response structure—concept / passage / graph / score— has a high degree of affinity with the reasoning model of generative AI.

2025-12-15

img-blur-shadow
Blog Semantic Integration Engine: A BoK Integration Demo for the AI Era

The Semantic Integration Engine rebuilds BoK knowledge into AI-interpretable semantic data by integrating RDF, vector search, and graph retrieval. This article explains the architecture, startup procedure, query example, and interpretation of the returned JSON.

2025-12-08