Shift Atlas

Role: Independent Builder
Scope: Knowledge System • Design Intelligence • Web
Impact: Project website in progress

000 // PROJECT OVERVIEW

Shift Atlas is a website for storing new design cognition. It is not a news feed or inspiration gallery. It is a structured archive that turns AI-era signals into judgments, topics, patterns, and frameworks for long-term design thinking.

001 // CHALLENGE

Most AI design websites stop at examples or pattern lists. They help you collect references, but they do not help you accumulate judgment. For long-term work on AI-era systems and human-AI relationships, I need a site that can preserve how a signal becomes a topic, how a topic becomes a pattern, and how repeated patterns become a framework.

002 // APPROACH

I am building the project as a knowledge system instead of a content site. The pipeline starts from source ingestion, then normalizes and scores signals, groups them into topics, classifies them as observe, update, or shift, and finally publishes them into topic dossiers and framework notes.

003 // SOLUTION

The website is structured around five content layers: signals, judgments, topics, patterns, and frameworks. It acts like a design intelligence atlas. The goal is to make new interaction paradigms, changing human-AI relationships, and long-term system-design insights visible and traceable over time.

004 // RESULTS

  • ▸Defined the first project architecture and knowledge model
  • ▸Confirmed the initial information-source pool and ingestion strategy
  • ▸Started the first web page for the project inside the portfolio

005 // CONFIRMED SOURCES

Initial information pool confirmed from the current ingest pipeline. The system mixes official, academic, media, and personal-analysis sources so judgments are not based on a single information surface.

Official

  • ▸Google DeepMind
  • ▸Google AI Blog
  • ▸Microsoft Research Blog
  • ▸OpenAI Newsroom via RSSHub + official fallback
  • ▸Anthropic News via RSSHub + official fallback

Academic / Research

  • ▸Hugging Face Blog
  • ▸Nielsen Norman Group
  • ▸Alignment Forum
  • ▸PaperWeekly
  • ▸vLLM Blog

Media

  • ▸The Verge AI
  • ▸MIT Technology Review
  • ▸TechCrunch AI
  • ▸VentureBeat AI
  • ▸Wired AI
  • ▸Ars Technica AI

Personal / Analysis

  • ▸Import AI
  • ▸AI Snake Oil
  • ▸Simon Willison
  • ▸Stratechery

006 // WEBSITE FRAMEWORK

LAYER_01

Signal Layer

Collect and normalize RSS, Atom, and official-page signals into a structured feed.

LAYER_02

Judgment Layer

Score and classify items as observe, update, or shift instead of forcing every signal into a strong claim.

LAYER_03

Topic Layer

Group multiple signals into long-running topics such as generative UI, AI search interfaces, or human-AI trust.

LAYER_04

Framework Layer

Promote repeated structural shifts into durable design principles and long-term methods.