New guide shows how to build a personal AI with user-owned memory

Author and former engineer Noboru Ikuta has released “AI of One’s Own,” a book, course and template pack that teaches non-programmers to build a personal AI environment on a Mac. The pitch: keep long-term memory under the user’s control even when the underlying cloud model changes.

Why it matters: - Current AI assistants often lose context between sessions, forcing users to repeat preferences, goals and past decisions. - Ikuta’s approach keeps the memory layer on the user’s own Mac, so that information can be read, edited, backed up, version-controlled and deleted by the user. - The model can still use cloud AI systems from OpenAI and Anthropic for intelligence, while the personal memory stays local.

What happened: - Noboru Ikuta released “AI of One’s Own,” a practical guide for building a personal AI environment with long-term memory under user control. - The release comes in three parts: a Kindle and paperback book, a screen-based Udemy video course and a configuration template pack on Gumroad. - The guide is available in English and Japanese. - More information and a free preview PDF are available at the official website.

The details: - The book centers on four structural pillars: system prompts, retrieval-augmented generation, persistent memory and pinned documents. - Ikuta wrote the guide for people with no programming experience. - Readers only need to be able to use a Mac, a browser and basic copy-and-paste terminal commands. - The memory layer includes documents, preferences and past decisions stored on the user’s Mac. - The book does not propose a fully local AI system. - Instead, it uses a hybrid setup that combines cloud models with local memory control. - Ikuta said the three formats are meant to cover thinking, hands-on building and shortcut templates.

Between the lines: - The book is aimed at a practical pain point, not a theoretical one: users want AI that remembers them without surrendering that memory to a platform. - The hybrid design suggests a middle path between fully cloud-based assistants and fully local AI tools. - Ikuta’s own background likely helps position the project as a builder’s guide rather than a pure consumer product. - Ikuta worked in telecommunications and database engineering, founded Infotect Inc. in 2004 and had green-IT research selected for Japan’s IPA Mitou Program in 2008. - He spent nearly seven years abroad from 2019 to 2026 as a single father raising three sons. - After returning to Japan, Ikuta built his own personal AI environment in about two weeks while managing school runs and household life, and that experience became the basis for the book.

What’s next: - Readers can use the free preview PDF before buying the book or taking the course. - The official site lists the synopsis, the four-pillar design and the memory chapter in the preview materials. - The broader test will be whether non-programmers can actually follow the workflow and maintain a durable personal AI setup over time.

The bottom line: - “AI of One’s Own” tries to make personal AI less like a disposable chat session and more like a user-controlled system with memory that survives model changes.