Intelligence Is Becoming Aluminum
A personal reflection on shrinking frontend roles, increasingly capable coding agents, the falling price of intelligence, and why changing careers may be changing cabins rather than finding a lifeboat.
AI
13 posts
Context Engineering Is Not About Adding More Context
A large context window only defines how much a model can see. Context engineering decides what enters that window, how to compress and order it, what to isolate, and what to discard as a task grows.
From One Model Call to an Agent: How LLMs, RAG, Tools, Loops, and ReAct Fit Together
LLMs, prompts, context, RAG, tools, workflows, loops, ReAct, memory, and agents belong to different layers. This guide explains what each one does and how they fit into a working system.
I Stopped Sending the Whole Book to the Model
My whole-book review feature used to send the setting, chapter plan, facts, and memories in one giant prompt. Two model calls took more than eight minutes. I replaced that request with a small, bounded agent loop that retrieves context only when it needs it.
Toolchain Performance Matters More Than Ever in the AI Era
AI coding is increasingly an agentic loop: write code, run validation, read failures, and fix again. In that loop, test, lint, and build performance directly shape engineering throughput.
The Next Step for Admin Platforms Is CLI and Skill
A reflection on why admin and configuration platforms should stop exposing their capabilities only through pages, and start turning operations into CLI commands and agent-readable skills.
gpt-image-2 Prompt Library: Technical Diagrams, Covers, and Visuals
Tested gpt-image-2 prompts for technical diagrams, infographics, editorial covers, miniature scenes, logos, and knowledge cards, organized by use case.
gpt-image-2 Prompts for Commercial Covers and Miniature Visuals
Tested gpt-image-2 prompts for editorial covers, miniature product scenes, technology posters, concept logos, and botanical knowledge cards.
gpt-image-2 Prompts for Technical Diagrams and Infographics
Tested gpt-image-2 prompts for turning architecture notes, technical mechanisms, and Mermaid flows into readable technical diagrams and infographics.
In the AI Era, Developers Have the Freedom to Refactor
A reflection on Bun's move toward Rust, a small SolidJS-to-Vue migration, and why technical choices in the AI era are becoming less like permanent bets and more like decisions that can be revised.
Why AI Has Higher Marginal Costs Than Internet Software
Marginal cost is the cost of serving one more unit. AI adds inference, tokens, GPU time, and tool calls to each request, changing software unit economics.
SillyTavern DeepSeek API Setup: Step-by-Step, No GPU Required
Configure the DeepSeek API in SillyTavern step by step with the built-in provider or an OpenAI-compatible connection, without a local GPU.
DeepSeek R1 with SillyTavern: Local Ollama Setup on Windows
A practical Windows setup for running DeepSeek R1 with Ollama and connecting SillyTavern to the local model service, covering model size, GPU requirements, API settings, and common problems.