Hey Reader,

A lot of the rules we built our work habits around made sense before AI. Some of them held up for decades. Right now, in early 2026, I think it's worth asking which ones still do.

Here are five things I think are worth throwing out:
​

1. Detailed mid-level planning

Knowing where you're going is still important. But the master plan that used to live between the destination and the work itself is mostly just a way to delay starting.

Before, mapping out how all the systems connect just felt like good planning. Getting things wrong mid-build was expensive, so you planned your way around that risk. But as I wrote earlier this month, speed is the only moat. The faster you move, the more AI labor you can put to use. A detailed plan sitting in the middle slows that down. Get clear on the destination, then get straight into the details.
​

2. Following whoever shows up in the feed

Curating your information sources is more important than ever.

Get rid of the content that isn’t serving you. Don’t just let the platforms suggest content and follow it blindly. I've spent time training the algorithms to only surface people sharing practical things. I don’t want to see the “world is ending" think pieces or the "look at the 37 agents I built" posts. I want to see people who are doing real things and showing their work.

I make a point to follow accounts with people who are doing the things I want to do. Like learning about hooks on Instagram.

Get really clear on your goals: what are you trying to accomplish in the next 3 months? Curate your feeds to align with that.
​

3. Learning before doing

You will learn faster by building something than by reading about it.

My son, August, was in the living room recently using Wispr Flow to describe what he wanted to make. He built a web-based video editor, got bored with it, and made a first-person shooter instead. He didn’t wonder whether he knew enough to start. The people I see moving fastest right now share that quality.
​

4. Staying in your lane

The idea that you should know your role and hire out everything else might have made sense when doing it yourself automatically meant doing it badly. But that’s no longer the case.

Dylan Feltus is a designer. He got frustrated with OpenClaw and didn't like that it was written in TypeScript, so he rebuilt it in Swift even though he's not an engineer. It's a cleaner, native Mac app one tenth of the original size. He just decided to do it. The question used to be whether something was in your skill set, but now it's more like whether you care enough to see it through.
​

5. Reading your code before it ships

For most of software history, everyone read their code before it shipped—which made sense when you were the one writing every line.

But front-end developers used to hand-code CSS, too. At least until Tailwind came along, and then nobody cared how it was written anymore. AI is changing things again. The code it generates may look a bit different than what you would have written yourself by hand, but as long as it works, and the architecture holds up, those are the important things.

—

Most of these things used to feel necessary. Some of them genuinely were. But really, they’re just different forms of waiting.

August didn't wait for a plan or anyone to tell him he could. He just built the thing.

You don't need permission to start.

​

PODCAST

Habits For (Almost) Limitless Energy

Simon Alexander Ong started his career as a finance analyst before becoming a speaker, coach, and author of the bestselling book, Energize.

In this episode, Simon shares his approach to sustaining energy across every dimension of life. He's got a lot of good thinking on why energy is harder to manage than time and how to get better at it.

We cover:

• Energy management versus time management
• The three keys towards taking action
• How to live a successful life

​​Watch or listen to episode »​​

KIT FEATURE

Trigger webhooks from Kit's Visual Automations

Kit's new webhook app lets you add a webhook action node to any Visual Automation. When a subscriber hits that step, Kit sends their data to whatever destination URL you configure, no custom code required.

Use it to connect Kit to platforms not in the App Store or trigger real-time workflows in tools like Zapier.

​Learn more »​

Don't have an account yet?

​Sign up for Kit »​

X POST

Former GitHub CEO's beliefs

Nat Friedman, former GitHub CEO, has a page on his personal website called "Some things I believe." Ben Lang shared it on X, and the full list is worth reading.

A few that stood out:

• A week is 2% of the year
• It's much easier to work on things that are exciting to you
• Great individuals should be fully empowered to exercise their judgment

​Read post »​

Have a great week!

—Nathan