After a few months of making AI teaching content, lining up a few of those old notes side by side actually traces out a pattern: what gets traffic and what doesn’t often doesn’t line up with what I personally think matters.
The Demand for the Lecture Is Real
A free lecture went from 200 to 350+ signups. Sales reps, PMs, marketers, researchers, designers, accountants, engineers, cram-school owners — every profession showed up. They’re all asking the same thing: “Has AI changed the workplace? Am I ready?”
The Job-Search Demo Got the Most Questions
Cut out the AI job-search demo segment from the 5/10 lecture and put it on YouTube: https://youtu.be/HcADayRCJMg. Claude Code customizes resumes for six companies in one pass, saving five evenings of manual work. This was the most-asked segment from the live lecture.
The Thing I Think Matters Most Pulled the Least
But turn the camera to the video I cared about most, and the numbers drop.
The advanced concept I personally consider most important — “on-demand loading and indexing” — actually pulls less view count. Could be I didn’t explain it plainly enough; could also be that even though everyone knows quota matters, the video didn’t aggressively tie context management to quota. Context management saves more than tokens; it raises model performance too.
The One Against the Wind
There’s another kind of low traffic — not a failure to explain it well, but shipping it straight into the wind on purpose.
When a wave of folks recently championed HTML over markdown, I shipped an honest video about my own Obsidian knowledge base. Does that count as going against the wind? https://youtu.be/EhMKfG1dvnI