AI insights, guides, and comparisons
Explore the latest articles on AI tools, workflow trends, product comparisons, and practical use cases.
'The model reads the whole paper' is a half-truth
People say Elicit and scite 'read' research. They read the abstract's angle and show you the cites.
The quiet winners of October's AI launches
A dated roundup of the AI launches that matter this week, minus the marketing.
The 'AI agent' pitch finally did something useful
For two years 'agent' meant a demo that booked a fake restaurant. Dots is the first one I'd point at a real chore.
I tried writing with the AI off for a day
One day, no ChatGPT in the draft. The copy was worse, then better.
The three-pass rule for any AI draft
Generate, cut, rewrite. Skip a pass and it reads like a brochure.
Gamma decks all look the same after a while
Great first draft, same ending. The template shows through if you don't fight it.
How I cite AI answers without lying to myself
An answer with a link is not a source. Here is the two-click rule I use.
Open weights still matter more than the demo
Mistral Large 4 promising open weights by month's end is bigger than its score.
I cloned my voice and felt weird about it
ElevenLabs v4 made a clone good enough to fool my own voicemail.
The cheap decision models are the real story of October
Forget the flagships. pplx-decider at $0.04 per million input is a different game.