Log In

The Export

Honest tips, hidden quirks, and practical insights about AI tools — from the people who use them all day.

AI Coding Agents Need A Preflight Context Budget
Before an AI coding agent reads half your repo, give it a context budget, file list, and receipt trail.
Read article →
The Hidden KVM Problem: Why My Monitor’s USB Ports Work With HDMI But Not DisplayPort
If DisplayPort video works but monitor USB dies, the KVM pairing may be sending your keyboard to the wrong computer.
Read article →
Why Everything Turns Into Two Sides, Even When It Isn’t
Why nature, psychology, politics and technology keep producing two opposing camps — and why the missing third option often matters most.
Read article →
The AI Coding CLI Is Now Part Of Your Supply Chain
AI coding CLIs can read files, call APIs, run commands, and sync context. Treat them like supply chain tools, not cute chat boxes.
Read article →
Are AI Assistants Replacing the Web Browser? What ChatGPT and Claude Can Actually Do on the Web
ChatGPT agent and Claude for Chrome let AI read pages and take actions, but neither replaces Chrome or Safari.
Read article →
Subscription Cancellation Is Now A Product Requirement
Click-to-cancel rules are turning cancellation UX into a product, legal, and logging requirement. Here is what teams should build.
Read article →
GPT-5.6 Makes The Model Switchboard Mandatory
GPT-5.6 is a reminder to route AI work by risk, cost, speed, and evidence instead of chasing one default model.
Read article →
Set Up Gmail For AI Agent Logins Without Handing Over Your Life
A practical Gmail setup for AI agent logins, aliases, labels, permissions, schedules, screenshots, and limits.
Read article →
Why Your Keyboard And Mouse Hate Your Monitor USB Hub
Monitor USB hubs are convenient until sleep, bandwidth, KVM, HDMI, and USB-C quirks make your keyboard vanish.
Read article →
GPT-Live Turns Voice AI Into The Interface
GPT-Live is a warning that voice AI needs receipts, context, and UI rules before it becomes the default interface.
Read article →
Which AI Is Best For Each Job, And The Quirk That Comes With It
A plain-English 2026 guide to which AI tool fits each job, from research and coding to office work, search, and daily planning.
Read article →
Local TTS Is The AI Feature Worth Shipping
Local text-to-speech is finally good enough for practical AI apps. Here is where it fits, where it fails, and how to ship it.
Read article →
AI Agents Need Office Files They Can Diff
AI agents can edit Office files, but Word, Excel, and PowerPoint workflows need diffs, previews, tests, and rollback.
Read article →
AI Agents Need Smaller Jobs, Not Bigger Promises
AI agents fail when they are given vague jobs. Here is how to scope, log, and supervise them without buying the fantasy.
Read article →
AI Travel Planning Still Needs Offline Maps
AI can draft a travel plan, but offline maps make it usable when signal, roaming, and app memory fail.
Read article →
Mechanical Turk’s Slow Fade Is A Data Quality Warning
Amazon put Mechanical Turk into maintenance. AI teams need better receipts for human review, labels, and data quality.
Read article →
What To Expect From The OpenAI And Anthropic IPOs
OpenAI and Anthropic IPOs may be huge, but investors will test compute costs, revenue quality, governance, and timing.
Read article →
AI Coding Agents Need Workspace Quarantine
AI coding agents can mix context through memory, folders, and tools. Treat every project like it needs quarantine.
Read article →