Notes on personal AI,
privacy, and hardware.
Trends, theses, and behind-the-scenes from the team building YeongSil — the autonomous personal AI device.
Why Personal AI Finally Needs a Body
Chatbots forget you. Smart speakers can't see. The next leap in AI isn't a bigger model — it's a device that lives with you, sees what you see, and acts on your behalf.
Read post →AI Privacy in 2026: Why On-Device Memory Matters
Cloud LLMs train on your prompts. Browser extensions leak your data. Here's how a device-first architecture changes the privacy equation for personal AI.
Read post →The 2027 AI Hardware Thesis: Why Investors Are Watching Standalone Devices
Smart glasses, AI pins, ambient assistants — the post-smartphone race is on. Here's the market thesis behind standalone, document-aware AI devices, and why the timing is now.
Read post →RAG vs Fine-tuning: What Your Personal AI Actually Uses
Fine-tuning was the answer when models were small and data was static. For a device that has to remember your lease, your prescriptions, and yesterday's invoice, RAG is the only architecture that holds up.
Read post →The Rabbit R1 Failed. Here's What We Learned.
The Rabbit R1 was the most-hyped AI gadget of 2024 and one of the most criticised by 2025. A fair post-mortem on what went wrong — and the three design decisions that shaped YeongSil in response.
Read post →The Humane AI Pin: A $700 Lesson in What AI Needs to Be
Humane raised $230 million, shipped a beautiful pin, and shut down hardware operations 18 months later. A fair look at what went wrong — and what every personal AI product should take from it.
Read post →What Makes a Personal AI Actually Personal?
ChatGPT is not personal. Siri is not personal. A model that knows the internet but not you is a stranger who happens to be articulate. Three properties — memory, presence, and action — separate personal AI from everything else.
Read post →YeongSil for Small Businesses: Your Staff Member That Never Forgets
A countertop AI device that has read every supplier invoice, every patient record, every booking sheet — and answers out loud, in the language you actually speak. Five concrete scenarios for shops, clinics, and hotels.
Read post →Privacy-First AI: What to Look for in 2026
Most AI products marketed as private are private in the marketing but not the architecture. Seven questions to ask before you hand a system your documents, your voice, or your face.
Read post →Building on YeongSil: What the Developer Platform Will Offer
A standalone AI device with a million-user installed base is a platform. The YeongSil developer programme, the Skill Store economics, and the categories that will define the first wave of third-party skills.
Read post →Ambient AI vs Smartphone AI: Why the Form Factor Changes Everything
Every AI use case that requires picking up your phone is bleeding attention. Ambient computing is not a nicer UI — it is the removal of the mode-switch cost that defines the smartphone era.
Read post →Stay synced with the YeongSil launch
Get new essays and product updates the moment they ship. No spam — just signal.
Join the waitlist →