← Back to siteWritingMuntahi Safwan
Writing / 9 articles

Notes, written slowly

LabWhy I Built Cloister: A 5 MB Messaging App in Rust and TauriCloister puts WhatsApp, Discord, Gmail and Messenger in one quiet window. Here is why I built it, how it stays small, and what I learned doing it in Rust.3 min read
11 November 2026AIHow to Make Your Website Readable by ChatGPT, Claude and PerplexityAI assistants now answer questions by reading websites. Here are the practical steps I took so they can read mine, and an honest note on what nobody can promise.3 min read
11 November 2026AIAI Agents That Draft, Humans Who Approve: How Octahive Is DesignedOctahive is agency software where AI agents do the routine drafting and a person approves anything that leaves. Here is the thinking behind that rule.3 min read
11 November 2026LabTauri vs Electron: What I Learned Shipping a Desktop AppI chose Tauri over Electron for Cloister and ended up with a 5 MB installer. Here is an honest comparison from someone who shipped with one of them.3 min read
11 November 2026StudioWhy I Rebuilt My Portfolio in Astro Instead of Next.jsMy old portfolio ran on Next.js. I rebuilt it in Astro, and on the way took the mobile Lighthouse score from 43 to 81. Here is what was slow and what fixed it.3 min read
11 November 2026LabWhat Is GraphRAG, and Why Does Compressing the Graph Matter?GraphRAG answers questions by walking a knowledge graph instead of only matching text. It is powerful and expensive. Here is how it works and why I am researching ways to shrink the graph.3 min read
11 November 2026AIWhy Every AI Answer Needs a Source: Notes on Building RAG ChatbotsA confident answer with nothing behind it is a guess. Here is why the first thing I add to an AI assistant is the part that shows where the answer came from.3 min read
11 November 2026AIHow Face Detection Works: Haar Cascades vs Deep LearningFace detection finds where the faces are in an image. Here is how the classic Haar cascade method and modern deep learning detectors work, with runnable code and advice on choosing between them.3 min read
8 August 2025AIHow Computer Vision Works: A Practical IntroductionComputer vision turns pixels into decisions. This is a practical overview of how it works, the ideas behind modern models, and a sensible way to start building.3 min read
8 August 2025