Danial Kalbasi

📍 Amsterdam http://danialkalbasi.com

Read a book in less than an hour

If you are into reading books, one of the best skills to master is the skimming technique. The idea is to grasp the book in under an hour and figure out whether reading cover to cover is worth your time. Skimming in a nutshell is essentially: * Read the title, headings,

Learning Dutch sucks more than learning English

One of my challenges with improving my Dutch is finding things to read that are both interesting and fit a beginner level. But it's extremely hard to find good resources considering two prerequisites. It also made me think why English is much more enjoyable to learn. When I

Devin gets it

Many ridiculed Devin when it came out, but in retrospect, the founders actually understood better than most where agents are heading — replacing an entire job, not merely improving it.

Dutch local times in 1909

Towns in the Netherlands used to have their own local time until 1909. So if you went from the east to the west, you would end up with 15min time difference. Interestingly when the Netherlands decided to unify the local times across the country, they didn't adopt GMT

Obsidian Vox: Listen to your notes with natural voices

I recently built a plugin for Obsidian to listen to my notes with natural voices that are generated by LLMs. It supports ElevenLabs, OpenAI and the built-in computer voice. The last one is not really natural, but convenient. I initially wanted to listen to a piece of writing in a

Future self in tradeoff analysis

A good technical tradeoff analysis comes down to whether your future self is willing to live with the solution. Say you are analyzing a problem and you have two solution options. Option A is technically superior to B. It's more novel, has new technical challenges that the team

More data is less

A pattern I've seen often is teams trying to prove an idea by asking for more data before they build anything. Data has two fundamental problems in this situation. One is that you can prove or kill a solution with the same exact data. It all depends on

Being logical or intuitive

I came to trust people's decisions when it comes from strong intuition rather than pure logic. Sometimes someone can sound logical in a discussion, but their advice is wrong if you wait long enough.

From prompting agents to building them

The way we build software with agents focuses too much on optimizing for efficiency per task rather than rethinking the entire workflow. Focusing on individual tasks creates busywork that is not worth spending one's time on. Consider reviewing AI-generated code or AI-generated documents that take far less time

Delegate to AI

Giving tasks to AI agents is similar to how we delegate to humans: - Delegate what you can't or won't learn from - Delegate repetition, routine, and noise - Do the work you do best - Skip when preparation exceeds value