Stay current on AI in 30 minutes a week
A weekly routine for keeping up with AI without drowning: what to read, what to ignore, and the one question that filters everything.
AI news arrives at a volume no working person can absorb, and most of it does not deserve absorbing. The skill is not reading faster. It is filtering harder. This guide is the routine: thirty minutes, once a week, with a filter that throws away ninety percent of the feed and keeps the part that affects your work.
The one-question filter
For every AI headline, ask: does this change what I can do, what it costs, or what I should stop trusting? Everything that matters to an operator lands in one of those three buckets.
A new model that handles longer documents: changes what you can do. A price drop or a capable cheaper tier: changes the cost calculus from the solo operator’s AI stack. A study showing a failure mode in a tool you rely on: changes what you trust, and updates your copy of what AI still gets wrong.
Everything else, and it is most things, fails the filter: funding rounds, executive moves, demo videos of products you cannot buy yet, benchmark wars between models you do not use, predictions about timelines nobody can verify. Interesting is not the bar. Actionable is the bar.
The thirty-minute shape
Put it on the calendar, same slot weekly. Ambient scrolling feels like staying informed and delivers anxiety instead; a bounded session delivers the same information without the cortisol.
Minutes one to ten: scan two or three digest sources, headlines only, marking what passes the one-question filter. A good weekly digest has already done the volume reduction for you, which is the entire reason digests exist. This site’s weekly brief is built to be exactly that for working readers.
Minutes ten to twenty-five: read the two or three items that passed, properly. Follow one to its primary source, the actual announcement or paper rather than the coverage of the coverage. Primary sources are shorter than the discourse about them surprisingly often.
Minutes twenty-five to thirty: write down, anywhere, one sentence per item: what changed and whether you will do anything. Most weeks the honest answer is “nothing yet,” and that is a finding, not a failure. The weeks that produce a “try this Monday” pay for the whole quarter of sessions.
Pick sources like you pick suppliers
Three or four sources, chosen deliberately, beat thirty tabs. A useful portfolio: one digest that summarizes the week, one practitioner who shares real hands-on results rather than takes, and one primary feed from the vendor whose tools you actually pay for. Diversity of type matters more than quantity, because the failure mode of a narrow feed is hearing the same enthusiasm three times and mistaking it for confirmation.
Audit the portfolio twice a year. A source that has not changed one of your decisions in six months is entertainment, and entertainment is fine, but it goes in the entertainment budget, not the thirty minutes.
Trying things without falling behind on work
Staying current includes occasionally touching the new thing, or the knowledge stays theoretical. Budget one experiment a month, an hour, on a real task from your own week, run like the pilot discipline in the automation playbook: a fair test, a real comparison, a keep-or-drop decision.
One honest experiment teaches more than twenty reviews, because it tests the tool against your work instead of against a demo script. And declining to chase the other twenty-nine new tools that month is not falling behind. It is the discipline that keeps the thirty minutes thirty.
The quiet advantage
The operators who do well with AI are rarely the ones who know the most news. They are the ones who reliably notice the four or five changes a year that matter for their business, and act on them while everyone else is still arguing about the other thousand headlines. A calendar slot, a hard filter, one experiment a month: that is the entire system, and it compounds.