|
||||||||||||
|
From the Desk
This week, AI started working without asking.
Two weeks ago AI started doing the work. This week it started doing it without asking.
Superhuman's email client now writes replies in your voice — and in testing, 60% went out without anyone editing them. Google turned NotebookLM into Gemini Notebook, which can run code inside your research. And the year's big AI safety report landed with a sobering headline: the best-scoring model earned a C+. Put those together and the job gets clearer. As AI does more on its own, the scarce skill isn't doing — it's judgment: knowing what to let run untouched and what you'd never send unread. Our first cohort proved they had it at Demo Day on Thursday.
|
||||||||||||
|
||||||||||||
|
02 · Try This Today
Draw your trust line.As AI starts sending things on its own, decide in advance what it may send, what it may draft, and what stays yours. Paste this into Claude, ChatGPT, or Gemini.
What you get: A clear, personal trust line — so the next time a tool offers to auto-send, you already know your answer.
|
||||||||||||
|
03 · Deep Dive
The more AI does on its own, the more your judgment is the product.
Last issue, AI crossed from answering to doing. This week it went further — doing the work unsupervised. Superhuman sends most of its AI drafts with no human touch. Meanwhile the year's safety report says the best model earns a C+. Both are true at once, and together they define the job.
When AI did nothing on its own, judgment was invisible — you did everything, so of course you checked it. Now that AI acts unwatched, judgment becomes the entire value you add: deciding what runs untouched, what you inspect, and what only you should do.
The operators winning right now aren't the ones who trust AI the most, or the least. They're the ones who've drawn a precise line — this is safe to automate, this needs my eyes, this is mine alone — because they understand the work well enough to know where the risk lives.
|
||||||||||||
|
04 · Tool Spotlight
Gamma.Describe the deck; it builds it.
gamma.app →
Gamma turns a prompt or a rough doc into a designed presentation, one-pager, or webpage in seconds — no fighting with slide layouts. For an operator or founder, it's the fastest path from "I need a deck for Thursday" to a real first draft you can refine.
The non-technical payoff: paste your messy notes, tell it who the audience is, and get a structured, on-brand draft you edit down — instead of staring at a blank slide at 11pm.
|
||||||||||||
|
05 · What We're Building
Demo Day happened. Cohort 2 is next.
On Thursday, Cohort 1 stood up in Austin and demoed the AI tools they built around their own jobs — operators, not engineers, showing things that actually run. That's the whole thesis, proven in a room. Cohort 2 starts August 3 — two weeks out, and the last cohort before fall.
Cohort 5 also available · Nov 30–Dec 11
|
||||||||||||
Know someone who should be reading this?Forward it to a founder or operator who needs the signal without the noise.
Forward this issue →
Free · Every Tuesday · makersquare.ai
|
||||||||||||
|
AI Builder Program · Austin, Texas
|