Selected articles

Worth reading

33 Questions Executives Ask About AI, Answered

Every (paywalled)Mike Taylor and Natalia QuinteroAugust 2026

Every's consulting team compiled the questions they hear most from executives and answered them in one piece. It covers strategy, governance, adoption resistance, and where to start.

Some of the answers are better than others, but as a map of what senior people are actually worried about when it comes to AI, it's the most useful single document I've seen.

If you're running a firm and wondering what your peers are asking behind closed doors, start here.

Key points from the article

  • Strategy. Most executives are asking the wrong first question ("which tools should we buy?") when they should be asking "which workflows are costing us the most time for the least judgement?"

  • Adoption resistance. The pattern is almost always middle management, not frontline staff, because middle managers correctly perceive that AI changes their role first.

  • Governance. The authors argue that waiting for perfect policies before starting is the most common and most expensive mistake.

  • Where to start. Pick one workflow, one team, run it for 30 days, then measure before and after.

  • My read. The article is strongest on the governance and adoption sections. The strategy advice is sound but generic.

The Biggest Tell That Something Was Written by AI

The Atlantic (metered paywall)Eve FairbanksMay 2026

Eve Fairbanks is a professional editor, and this piece is the best articulation I've read of why AI-generated writing falls apart under scrutiny.

Her argument isn't about em dashes or word choice (though she covers those). It's that writing is thinking, and AI doesn't think: it produces grammatically perfect text with no underlying reasoning.

That matters for anyone using AI to draft client-facing documents. The output reads well until someone interrogates it, and then every part is equally wrong. This is exactly why Maneform's approach keeps the human in the drafting loop rather than handing it to a prompt.

Key points from the article

  • The surface tells. Em dashes, colons, and "it's not X, it's Y" constructions are easy to strip out. Tutorials already exist for doing so.

  • The deeper problem. Writing is a thinking process. When a sentence won't come right, your mind is often telling you the idea itself needs rethinking, or the message shouldn't be sent at all. AI skips that process entirely.

  • Why editing fails. Fairbanks describes editing AI text as trying to operate on a body where every organ is compromised: there's nothing to leave intact and nowhere to begin.

  • Her test. She asked ChatGPT to justify its own metaphor choices. It generated increasingly bizarre explanations ("raccoons have logistical agency") without ever recognising that the metaphor didn't work.

  • The implication. The output looks polished until a reader pushes on any part of it, and then the whole thing unravels.