Nate's Substack: the five human skills that grow alongside AI agents

By Nate from Nate's Substack · October 11, 2026.

Nate's Sunday briefing starts from a personal observation: the author says he is seeing "personal software" everywhere. He recounts that people around him are building small applications that fit some specific part of their work or their daily life, and although he does not have permission to detail those stories, he says the pattern has become impossible to ignore.

According to Nate, that activity has been "supercharged" since the arrival of Muse, which he describes as Meta's personal AI agent, built to carry out tasks through the browser and connected apps. He notes that, alongside this wave of app building, something else is happening: agents are starting to take over the ground people used to cover by "vibe coding" small applications of their own.

What interests Nate is not so much the tool as the effect it has on the person: whoever builds something knows what they want because they have lived with the problem, and now they have the means to act on it. He argues that AI has expanded what we are capable of doing faster than many people have changed their willingness to try; but when someone does try, they can learn to become more capable, and what they end up carrying out shapes what they will be willing to attempt next.

That, the author says, is where five skills converge: taste gives a reason to act; agency puts that taste to work before you know everything the task will demand; keeping focus exposes you to the hard parts of the problem; resourcefulness helps you learn to get through them; and follow-through lets you discover what actually helps the person you were building for. These skills, Nate maintains, are developed through the work itself, and in turn are what allow you to take on more and more work.

Nate adds that your sense of what you can achieve comes partly from accumulated experience: which ideas are feasible, which require outside help and which fall so far beyond your own skills or budget that they are barely considered. Those judgments, he says, help you function day to day, but they tend to persist even after the circumstances that produced them have changed.

The email previews that the full briefing (available at the link to the post) develops five sections:

- Staying with the problem as the approach changes: how to move between a personal app, an agent and agent-written code without losing what you learned about the original problem.

- Letting a first attempt sharpen your taste: how an imperfect attempt surfaces information that more extensive preparation might never produce.

- Growing your own agency through a real project: what a Reuters journalist learned building an AI tutor for her son, and how that project expanded what she was able to take on afterward.

- Giving focus time to pay off: why AI makes sustained attention more decisive, illustrated by what the solo developer of Stardew Valley learned when he removed a game system that, in principle, worked.

- A "prompt" for the shelved project: designed to decide whether to resume, reframe or definitively set aside a project you or your team dropped at some point.

The body of the email is the introduction/summary of this briefing and does not develop the full content of each section (that development is left to the linked full post), so this summary is limited to what the email itself states explicitly, without adding unmentioned details about the cases cited.