Started Development of Personal AI Agent MAHORO

On April 15, 2026, I officially kicked off development of MAHORO (Memory-Aware Hybrid-Ontology Reflective Oracle) — the core agent of MAiD, a personal AI agent project I have been working toward for years.

MAiD stands for Multi Agentic identity-Daemon. It is a Neuro-Symbolic AI personal agent framework that combines a self-implemented AIML 2.0-based natural language dialogue engine with LLM/Agentic RAG and Web ontologies for cooperative knowledge processing. A planned extension, CTML with NeSy, will add non-deterministic semantic checking to CTML (Conversation Test Markup Language), our own dialogue testing framework. Built entirely in Python. The name MAHORO comes from the ancient Japanese word mahoroba — meaning an ideal land, a place of peace.

The core concept of MAHORO is “the feeling of being understood.” Rather than a task-executing assistant, MAHORO is designed to grow into an agent that is yours alone — remembering your past experiences, choices, and feelings, and deepening its understanding the more you interact with it. The memories you accumulate make MAHORO uniquely yours — this is the idea at the heart of the design.

Technically, the architecture combines an AIML engine handling natural language processing and dialogue flexibility, a Web ontology layer handling logical reasoning, and an LLM providing intuitive knowledge supplementation — all working in cooperation. As of now, the foundational implementation of the AIML engine, memory system, and LLM fallback is complete, with CTML tests passing at 401/401.

Here are a few screenshots of the current behavior (UI is in Japanese):

This is a dream project I have been chasing since I was 20. I will keep sharing updates on the development here.