Peyman

AI writes in everyone's voice but yours. Well... building the fix.

AI tokenmaxxing is good and all, but... Have you tried a good night sleep? It's really underrated. Totally recommend.

This week, we're shipping a huge upgrade to Laravel Managed Queues. Cold starts from 30s -> 500ms, simplified configuration, FIFO support, and up to 1 hour job duration. Managed Queues scale to zero (so you pay nothing) when there are no jobs on your queue and scale up in milliseconds when needed. Unbelievable DX and work by the team. Absolute no brainer to enable Managed Queues for every project that uses queues on Laravel Cloud.

A small codex workflow that does WONDER, at almost no cost Ask GPT 5.6 Luna (Max) to use the ImageGen skill for a 3D low-poly object. Then to use the img2threejs skill. Tada. I didn't even spend 1% of my weekly tokens for this.

@a3b0_ @krzyzanowskim ProseMirror is definitely a good starting point, and I agree, agents can get 90% one shot. but I think we need better wrappers for the number of projects that now rely on these things. It's a big market

Is there anything more difficult than building a text editor?

@krzyzanowskim editors

@iamp3yman editors or cancer?

@krzyzanowskim yeah, but not enough people working on fixing this

@iamp3yman curing cancer is pretty close

@asyncze I really, really hope those rockets do not need a text editor on top of their usual challenges😅

@iamp3yman Building rockets is hard too

Building Neveshte, I look at AI that always returns an answer sometimes it should stop: • promise conflicts with evidence • audience is unclear • source can’t support the claim Generating while guessing feels helpful, but exposing the unresolved is better. Isn't it?

Before starting the next stage, ask: What is decided? What is still uncertain? What evidence would resolve it? Which downstream work depends on it? Creativity needs discovery. Put discovery in the cheapest stage.Content creation

AI makes late decisions look cheaper. It can generate 10 hooks, 3 scripts and 20 clips around an unresolved idea. The assets look finished. The ambiguity is still there. AI doesn’t remove decision debt. It packages it.

The most expensive creative decision is the one made one stage too late. Finding the angle during scripting. Finding the story while filming. Finding the structure in the edit. Every late decision turns creative work into rework.

Insane numbers!

Batch the labor. Don’t batch the hypothesis. 10 thumbnail exports: efficient. 10 videos built around an untested promise: expensive. Anything that needs audience feedback should stay small and easy to change. Am I wrong?

Guys, if you need an onboarding tour to explain your UI, the problem is the UI. Not the user.

Codex down? I just wanted to make a PR and a Linear issue before leaving the house and not thinking about coding for a few hours 😅

@mnowakdesign Just beautiful. It's been a while since I saw a design this exciting and calming at the same time.

some cinematic shots from the new dashboard animation 🎥

Before asking AI to rewrite a paragraph, label three things: Claim: What am I saying? Evidence: Why should anyone believe it? Consequence: Why does it matter? If the rewrite keeps the claim but drops the evidence or consequence, it becomes smoother, not stronger.

Come ship with us. 🫶