Notes from the workshop.
Practical writing on AI, product engineering and the decisions behind the things we ship — from the studio that builds six products of its own.
-
GEO: How to Get Your Product Recommended by ChatGPT, Claude and Perplexity
Generative engine optimization is how you get named when an AI assistant answers "what tool should I use?" Here is the practical checklist we run on every product we ship.
-
Your Best AI Answers Are Buried in Chat History (and You Will Never Find Them Again)
You have solved the same problem with ChatGPT three times because you could not find the answer you got last month. Here is why AI chat history fails as a knowledge base — and what to do instead.
-
AI Agents for Small Businesses: What Actually Works (and What Wastes Money)
Most small-business AI agent projects fail for the same three reasons. Here are the automations that reliably pay for themselves, the ones that do not, and how to tell them apart before you spend.
-
How to Ship an MVP in One Week (Without Shipping Junk)
A one-week MVP is not a rushed product — it is a ruthlessly scoped one. Here is the exact day-by-day process we use, and the seven things we cut every single time.
-
Prompt Engineering That Ships: 8 Patterns We Use in Production
Prompt tricks from Twitter threads fall apart under real traffic. These are the eight patterns that survived contact with production systems we actually maintain.
-
Why You Snooze Through Alarms — and What Actually Gets You Out of Bed
Hitting snooze is not a willpower problem, it is a design problem. Here is what sleep inertia actually does to your brain, and why an alarm you have to solve beats an alarm you can silence.
-
llms.txt: What It Is, Why It Matters, and How to Write a Good One
llms.txt is a plain-text file that tells AI assistants exactly what your business does. Here is what belongs in it, what does not, and the template we use on every site we ship.
-
Every Free PDF Tool Wants Your File. Ours Never Gets It.
When you split a contract on a free PDF website, that contract is uploaded to a server you know nothing about. It does not have to work that way — modern browsers can do the whole job locally.
-
RAG or Fine-Tuning? A Decision Guide for People Who Have to Ship
RAG teaches a model what you know. Fine-tuning teaches it how you behave. Most teams reach for the wrong one — here is the decision tree we use, and the cheaper option most projects overlook.
-
The QR Code on Your Menu Is Rented. Nobody Told You That.
Thousands of printed QR codes stop working every month — not because the technology failed, but because a subscription lapsed. Here is how QR services actually work, and how to avoid the trap.
-
How We Choose a Tech Stack in 2026 (and Why It Is Usually Boring)
The stack that gets you to launch is rarely the interesting one. Here is the decision framework we use on client projects, and the defaults we reach for when nothing forces our hand.
-
The Notebook Problem: Why Tracking Home Staff Attendance Always Ends in an Argument
Half-days, leaves, advances, overtime — and a memory-based record on both sides. Here is why domestic staff salary calculations go wrong every month, and the fix that takes thirty seconds a day.
-
How to Use AI Coding Assistants Without Wrecking Your Codebase
AI writes code fast. It also writes plausible code that quietly does the wrong thing. Here is the working discipline we use to get the speed without the six-month cleanup.
-
Cloudflare vs Vercel: How We Actually Decide Where to Deploy
Both are excellent. They are good at different things, and the wrong choice shows up as either a surprise bill or a week of fighting the runtime. Here is the split we use.
-
One-Time Pricing vs Subscriptions: What We Learned Shipping Both
Subscription is the default advice and it is often wrong. Here is the test we use to decide, and why two of our own products charge once and never again.
No posts in this category yet.