Archive
Observations
Less of a filing cabinet, more of a graph. Jump in by concept, follow the lines, open a cup when something catches.
Network
Each cup is a piece. Lines appear when two pieces share a lane. Filters highlight, they don’t hide.
Chronology
All cups, newest first
-
Cup 28 ·
The Companies People Want Cloned
SAASpocalypse vibes: reverse-engineering products, cloning software getting easy, and what that means for the industry.
Industry shifts Agents
-
Cup 27 ·
What an Agentic OS Actually Needs
What an agentic OS would need beyond a chat box with desktop wallpaper.
Agents Local & homelab
-
Cup 26 ·
Your Notes Know Something You Don't
Organizing notes so future-you (and your models) can find the thread: graphs, links, context.
Context & memory Human side
-
Cup 25 ·
The AI Revolution Runs on a Format From 2004
Markdown as the boring format the AI stack still runs on, from prompts to notes.
Context & memory Tooling
-
Cup 24 ·
The 99-Cent Flashlight App of AI
The thin apps wrapped around models: harnesses, flashlight-app energy, and who captures the value.
Agents Tooling
-
Cup 23 ·
The Hidden Cost of Looking Smarter
Why looking smart on benchmarks can be expensive theater, and what the scores miss.
Models & compute Agents
-
Cup 22 ·
The Test AI Already Passed
AI in classrooms: cheating panic, real learning, and the boring middle path that might work.
Human side Agents
-
Cup 21 ·
The Future of Webpages and the YOLO AI Strategy
What pages are for when agents and AI search read them more than humans do.
The changing web Agents
-
Cup 20 ·
The End of the Adapter Drawer
APIs and MCP as the USB-C moment for AI tools: one way models plug into the rest of the world.
Tooling Agents
-
Cup 19 ·
Does Originality Still Matter?
Watermarks, originality, and whether AI-made content still counts when everything looks generated.
Industry shifts The changing web
-
Cup 18 ·
AI Agents: What Does That Even Mean?
What people mean by AI agents: who decides the next step, and why the word is so slippery.
Agents Tooling
-
Cup 17 ·
The Invisible Web: How AI Reads Past the Spin
AI reading past your marketing spin: the invisible web and what it averages about your company.
The changing web Industry shifts
-
Cup 16 ·
How Far Are We From Dead Internet?
How much of the net is bots and synthetic sludge, and what that does to trust online.
The changing web Industry shifts
-
Cup 15 ·
Your Old Laptop Wants A Job
Giving an old laptop a job: self-hosting, local models, and owning a slice of your stack.
Local & homelab Models & compute
-
Cup 14 ·
Hail to the Algorithms!
How ranking systems quietly steer jobs, feeds, and choices, and what oversight even means.
Industry shifts The changing web
-
Cup 13 ·
You Don't Need a Death Star to Make Toast
Small language models versus always waking a giant one: when bigger is the wrong tool.
Models & compute Local & homelab
-
Cup 12 ·
What If the Lazy Programmers Were Right?
AI coding as the next dependency: trust, laziness as a virtue, and why open source felt the same once.
Agents Human side
-
Cup 11 ·
Why My AI Talks to Me Like a Caveman
Tokens as the weird little units models think in, and why they change how you write prompts.
Models & compute Prompting
-
Cup 10 ·
If a Company Can't Organize Its Website, What Did You Expect From Its Data?
Why messy websites and messy data make enterprise AI demos fall over.
Industry shifts The changing web
-
Cup 09 ·
Local vs Cloud AI: Who Owns Your Thinking?
Who owns your thinking: running models at home versus sending everything to the cloud.
Local & homelab Models & compute
-
Cup 08 ·
RAG vs Long Context: How AI Remembers Things
RAG versus stuffing the whole window: two ways models remember, and when each fits.
Context & memory Models & compute
-
Cup 07 ·
Where Does the Heat Go When You Press Enter?
What actually happens when you press enter: tokens, energy, and the cost behind a reply.
Models & compute Prompting
-
Cup 06 ·
Why I switched to Linux after 25 years
Switching to Linux after decades, and why local control fits an AI-heavy desk.
Local & homelab Tooling
-
Cup 05 ·
Context Beats Prompting
Context beats clever prompting: what you put around the model matters more than the magic words.
Context & memory Prompting
-
Cup 04 ·
Are Introverts Better at Working with AI?
Whether quiet, written thinking is a natural fit for chat-with-a-model work styles.
Human side Prompting
-
Cup 03 ·
The Renaissance of CLI
Why the command line is having a moment again in an AI-shaped workflow.
Tooling Local & homelab
-
Cup 02 ·
The Prompt Is the New Interface
Prompts as the new UI: how talking to models replaces menus, and why the CLI is suddenly cool again.
Prompting Tooling
-
Cup 01 ·
Learning has never been easier. That's the problem.
Using AI to study without turning learning into content production, and why simple tools win.
Human side Prompting