ipixxTalk to us

Journal / AI in marketing: the owner's field guide / What AI should (and shouldn't) do in your marketing

What AI should (and shouldn't) do in your marketing

A clear division of labor beats both extremes — all-human and all-AI each fail for opposite reasons.

Published · Updated · 5 min read · by ipixx

A precision robotic arm and worn wooden hand tools side by side on a dark workbench

Every argument about AI in marketing comes down to one question: which jobs is it doing? Get the division of labor right and the tool disappears into the work. Get it wrong and you either pay for slow manual labor or publish slop.

Give AI the reps

  • First drafts and variants — the blank page is its whole reason to exist.
  • Repurposing: one good idea, correctly reshaped for web, email, social.
  • Research summaries and competitive sweeps a person then verifies.
  • Formatting, consistency, alt text, metadata — the chores that eat weeks.

Keep people on the calls

  • What's true — claims, prices, promises, anything a customer could hold you to.
  • What's you — voice, taste, and the opinions that make anyone care.
  • What publishes — a person reviews it and the owner approves it, every time.
  • Where money moves — spend, sends, and anything touching a live system.

The failure modes, both directions

Doing everything by hand can mean paying for hours a tool could have saved. Handing everything to a tool can mean publishing a lot of nothing. The approach that holds up is unglamorous: the tool does the reps, people make the calls, the owner gives the yes.

See how we plan the work, prepare it, and ask for your approval before anything goes live.