Blog

The Complete Guide to Computer Specs for AI: From Cloud to Local LLMs (Windows & Mac)

What computer do you actually need to run AI and LLMs? Cloud AI like Claude and ChatGPT needs no graphics card, just 32GB of RAM. Running models locally is a different story: a usable 32B model needs 24GB of VRAM or 48GB of Mac unified memory, and 70B and up realistically needs 128GB on a Mac.

When AI Can't Render Korean Text: Background from AI, Text from Code

AI image generators break Korean text, logos, and exact pixel dimensions. The fix: generate only the background scene with AI (Claude delegates to Codex imagegen), then layer Korean text and logos via HTML with web fonts, captured to PNG with a headless browser. A build log for automating blog illustrations and OG cards.

Why the 1200px Card You Made Became an Unreadable Blob in KakaoTalk

Why a carefully crafted OG image turns into a blurry smear in KakaoTalk (Korea's dominant messenger app). The 9px math of canvas vs. render, how gaze-cueing works, URL cache invalidation, and why 7:1 WCAG contrast is the real minimum, all traced back to primary sources.

How Christianity Spread: Dissecting History's Most Successful Growth Marketing

A case study dissecting how early Christianity grew from roughly 1,000 followers to 35 million over 350 years, analyzed through nine growth marketing levers: addressable market, value, price, channel, promotion, segmentation, people, process, and growth engine. Historical figures are academic estimates.

The First Button of Any AI Marketing Agent: A Brand Brief, Not a Smarter Model

A brand brief is the single shared input for market-research, positioning, growth, and content agents. Without it, every agent fills in its own assumptions which then contradict one another. The secret to a good brief is honestly marking unknown fields as TBD rather than guessing. Includes a 10-section template and a real example.

Making Hundreds of Notes Fast for AI: I Tried Ontology, Then Found Something Lighter

Hundreds of notes piled up but AI still had to open every file to answer a simple question. I thought a full ontology would fix it. A lightweight index layer built from existing frontmatter turned out to work 2.9-13.3x cheaper. This is the build log.

I Logged In and Every User Property Was undefined: GTM and GA4 from Setup to Revenue Analysis

Can You Still Retarget Users Who Withdrew Marketing Consent: A Pixel Gating Design Story

How to Make Your Ads Learn Revenue: An Offline Conversion Tracking Case Study

Kakao Share Thumbnail Not Showing? Here's the Fix (JavaScript SDK 2025)

Learn why Kakao Share thumbnails don't appear even with a valid imageUrl, and fix it using scrapImage to convert your URL into a KAGE CDN URL. Step-by-step guide.

How to Generate CRM User Manuals in 20 Minutes with Claude Code Sub-Agents

Learn how Claude Code sub-agents can analyze your CRM codebase and auto-generate user manuals and visual guidebooks in just 20 minutes. See the real-world results.

GEO Generative Engine Optimization: The Complete 2025 AI Marketing Strategy Guide

Learn what GEO (Generative Engine Optimization) is, how it differs from traditional SEO, and discover 5 proven AI marketing strategies to get cited by AI search engines in 2025.