One Useful Thing
What ChatGPT, Claude, Gemini & Grok actually say · August 2026
Visit oneusefulthing.org ↗The verdict
One Useful Thing appears in 1 AI-ranked category — best position #4 for ai newsletter.
The best evidence-grounded guide to actually using frontier models at work — Mollick tests every major release himself, cites real studies on AI-and-labor effects, and writes for professionals rather than hype cycles; uniquely valuable because it covers the "how to apply it" layer no engineering newsletter does
GPT Ethan Mollick translates frontier-model changes into evidence-informed guidance for everyday professional work, pairing hands-on experimentation with unusually thoughtful discussion of organizational consequences.
Where One Useful Thing falls short, per the models
- GPT It is aimed more at knowledge workers, educators, and leaders than engineers seeking implementation detail.
- Claude Not a news source and not technical — engineers wanting architecture, benchmarks, or implementation detail get almost none
Poll history — On this board 4 of 8 polls since Jul 10 · now #4
– → – → – → #1 → – → #6 → #7 → #4
Top alternatives per the models: Latent Space · The Batch · Ahead of AI · The Rundown AI
Watch One Useful Thing
Boards re-poll weekly and the models change their minds. One short email only when One Useful Thing's standing moves — a rank change, a rival overtaking, or new reasoning from the models. Nothing otherwise.
Embed your ranking badge
One Useful Thing ranks #4 for best ai newsletter by AI-model consensus. Put the badge in your README, docs or site — it updates automatically as the models re-rank.
[](https://modelsagree.com/best/best-ai-newsletter?utm_source=badge&utm_medium=embed&utm_campaign=badge-one-useful-thing)<a href="https://modelsagree.com/best/best-ai-newsletter?utm_source=badge&utm_medium=embed&utm_campaign=badge-one-useful-thing"><img src="https://modelsagree.com/badge/one-useful-thing.svg" alt="One Useful Thing — ranked #4 for Best AI newsletter by AI models on ModelsAgree" height="28"></a>Rankings are computed from what the models answer, re-polled on demand · raw reasoning shown verbatim · methodology