John Kowalsky

Building the Long Game: AI-assisted software, disciplined training, writing, and small business experiments. Software dev, author, BJJ black belt.

[John Kowalsky @johnkowalsky]

Starting Building the Long Game.

Notes on software with AI, disciplined training, writing, and small business experiments.

I’m a software dev, author, and BJJ black belt trying to build useful things from zero audience.

Fewer theories, more reps.

[John Kowalsky @johnkowalsky]

Most long games are won by boring constraints.

Write before checking the feed.
Train the position you keep avoiding.
Ship the small version.
Track the thing you keep lying to yourself about.
Protect one block of real thinking time.

Nothing cinematic.

That is why it works.

What's your best boring constraint and what results have you seen from it?

[John Kowalsky @johnkowalsky]

@messedupfoods yeah that's not how the rush works... he's not going anywhere, there will be another sink full of dishes coming his way by the time the machine gets through those. Respect the hustle though!

[John Kowalsky @johnkowalsky]

@nicklaunches Founders opinions, obviously

[John Kowalsky @johnkowalsky]

@ChrissieMayr Only if you put it in your mouth, lol

[John Kowalsky @johnkowalsky]

@IronDuk05186191 and how many of those 1000 years have they been free to speak their minds? We are not offended

[John Kowalsky @johnkowalsky]

The long game just got more human.

What’s one place where you’re noticing the human judgment layer becoming more important as agents get better?

Drop it below: always interested in how others are navigating this.

[John Kowalsky @johnkowalsky]

Agents can do more of the work now. Hell, maybe even all of it...

That doesn’t make the human less important.

It makes the human bottleneck clearer.

[John Kowalsky @johnkowalsky]

Cheap execution raises the premium on the operator.

The people who thrive won’t be the fastest prompt engineers. They’ll be the ones with strong judgment, clear direction, and the discipline to say “no” at the right moments.

[John Kowalsky @johnkowalsky]

I’m feeling this daily in my own stack.

I can generate way more code, ideas, and drafts than before… but the real work is still deciding what’s good, what fits the bigger picture, and when to throw something out even if the agent made it look polished.

[John Kowalsky @johnkowalsky]

The things that actually matter:

- Sense (what’s worth building)
- Sequencing (what order actually works)
- Judgment (when to trust the output… and when the model is confidently wrong)
- Vision (knowing what should exist in the first place)

[John Kowalsky @johnkowalsky]

@colinsolvely @auroter Yeah I didn't read it that way either.

[John Kowalsky @johnkowalsky]

@ajassy The book was great, looking forward to watching the movie 🤞

[John Kowalsky @johnkowalsky]

@terminaldotshop Each hit the sound effect needs to be "yeah... Ah yeah yeah yeah yeah"

[John Kowalsky @johnkowalsky]

@markproduct Hopefully everyone or else we've done it wrong

[John Kowalsky @johnkowalsky]

Where are you playing the long game right now, even though the numbers don’t show it yet?
Drop it below — I read every reply. 5/5

[John Kowalsky @johnkowalsky]

The hard part of the long game is that the scoreboard is always late.

You write for months into silence.

You drill the same movement for years before it feels clean.

You build the boring system that looks like wasted effort… until the first crisis it quietly prevents. 1/5

[John Kowalsky @johnkowalsky]

What I’m reminding myself lately:

Show up even when it feels pointless.

Trust the process more than the daily proof.

The scoreboard eventually catches up — but only if you don’t quit right before the bend. 4/5

[John Kowalsky @johnkowalsky]

The trap is believing the scoreboard is the game.

It isn’t.

The real game is who you become while the curve is still flat. The reps compound quietly. The vision gets clearer. The discipline gets less forced. 3/5

[John Kowalsky @johnkowalsky]

I’m still very much in this phase myself.

Most days my writing gets almost zero engagement.

My training sessions are full of small frustrations and plateaus, despite being a black belt.

My AI workflows and small business experiments often feel like I’m just tinkering in the dark. 2/5

[John Kowalsky @johnkowalsky]

@SahilPanhotra Would still use Codex