Direct answer: stop trying to follow AI news and maintain a system instead: a weekly 30-minute read of three to five curated sources, one monthly bench test of a new capability on your own work, a quarterly hour re-checking what AI can now do that it couldn't, and a short monthly note shared with your team. Currency is knowing whether your working knowledge still holds, not knowing what happened today.

Nobody keeps up with the feed. Thousands of tools launch monthly and the commentary industry produces AI content faster than AI produces everything else. The people who appear current are curating harder than you, or performing.

Why "keeping up" is the wrong goal

Professional skills now have a shrinking half-life (World Economic Forum and IBM analyses put technical skills at just a few years), and AI-adjacent knowledge moves fastest. But the practical question was never "did you see the launch?" It's "do your working assumptions still hold?" That's a finite maintenance problem, and finite problems can be scheduled.

The weekly habit: a source diet, not a feed

Three to five sources, chosen by criteria: evidence over takes (do they cite checkable things?), vendor-neutral over vendor-fed, practitioner-relevant over researcher-complete. Read them batched, in one 30-minute slot, with one question in hand: does anything here change what I do? Then, honestly the highest-value step: unsubscribe from everything that only ever makes you feel behind. Fear-based AI content is a genre with a business model. You're not required to fund it.

The monthly habit: testing beats reading

One new capability, bench-tested on your own real task, verdict written down. One hour. Aim it at the edge: things that failed six months ago, or capabilities the sources claim just crossed over, because a crossing changes what you should delegate. Twelve written tests a year is more verified, personal AI knowledge than most marketers will ever hold.

The quarterly habit: re-date everything

One scheduled hour: walk your task map (what can AI now do?), your tool stack (what earned its subscription last month?), and your saved templates (which quietly stopped working?). Staleness in an AI practice looks exactly like smoothness, right up until someone with fresher assumptions does in one session what your setup does in three.

The multiplier: share a three-line note

Monthly, to your team: what I tested and the verdict, one change that matters for us, what I'm watching. Ten minutes to write. It enforces the other habits, deepens your own learning, and quietly makes you the building's calm authority on AI, which is a title nobody grants formally and everybody knows who holds.

Sign up to Marketing and AI: Foundations

FAQ

What are the best sources for AI marketing news? Judge candidates by criteria rather than names: cited evidence, vendor neutrality, practitioner relevance, and a calm publishing rhythm. Three to five is enough; more is a feed again.

How often should marketers test new AI tools? Once a month, deliberately, on a real task, with a written verdict. Less and your assumptions fossilise; much more and you've become a full-time benchmarker.

Is it too late to start learning AI for marketing? No. The field is a few years old, the deep practice layers are still forming, and a systematic starter overtakes a panicked veteran within months.