-
Here is a strange issue in Photomator: during photo editing, an .XMP file is created, which likely contains the editing steps applied to the photo. Additionally, a Photomator native file is also generated. Some of these files are very small, while others are quite large. For example, one file is less than 300 KB, while another is nearly 120 MB. Despite using the same editing process, the file sizes vary significantly. These files aren’t package files, it’s impossible to peek inside them. This is strange.
-
As a blogger and someone curious about many topics, I often feel I miss opportunities to become an expert in certain fields. Generative AI is one example. It gained public attention in the fall of 2022. At that time, when generative AI started to gather attention in the public, I should have recognized this important moment. I should have taken the chance to gather knowledge and organize it like a true researcher. I’m very analytical in general. I ask good questions. I have the tools and the motivation to do this work. Now, it feels overwhelming to catch up.
-
I’m spending a lot of time online and on my computer each week for my blogs, probably around 10 to 25 hours. I wonder how skilled I could become if I dedicated those hours to learning a new field like psychology or music.
-
Something Is Wrong
We’re getting close to the end of August and moving into September, which means Apple’s usual rush of new product launches. We already know quite a bit about Apple’s upcoming iPhone. We also know that Apple’s unifying Liquid Glass is likely to cause major issues when it encounters hundreds of millions of devices. We can probably expect Apple to be in damage control again. And I’ll probably have to explain why Apple did this to my friends and colleagues. Continue reading →
-
Out of curiosity, are you considering getting more RAM with your next computer to run local LLMs in the future?
-
If you have digital assets on GitHub, if your software dev workflows depend on GitHub, you might want to read this. Should we trust Microsoft, now?
-
AI as a Complementary Tool
One of my rules for using generative AI is to avoid starting a creative project with AI prompts. Today, someone at the office challenged this rule, arguing that beginning with AI can effectively kick-start the creative journey. While that may be true, I prefer to keep the human touch at the center and use AI as a complement to the process. Starting with AI risks making us lazy in the long run. Continue reading →
-
Fun fact: I keep a log (or a journal) of everything I do in a day at work. I’ve been doing this for decades. I always used digital tools for that but the guy who gave me the inspiration for doing this when I started my career was doing the same but on paper. I thought Incould improve his process. I did.
-
When creating backlinks in apps like Craft, if there is no automatic suggestions based on current document’s content, as the number of documents grows, it becomes mostly impossible to select related documents to potentially link to.
-
I began to unpublish several posts from numericcitizen.me, mainly those imported from Substack. It’s fascinating how quickly content can become irrelevant in our rapidly changing world.
-
Learning to use Confluence when you already know a lot about Notion feels like a major step back. 🥴
-
If Liquid Glass is about unifying all Apple’s platforms, why is it so toned down on the Mac? Asking for a friend.
-
Do you remember the SETI program? I think this is what it was called. Anyway, it was about providing or contributing personal computing resources to help find extraterrestrial life from background noises or signals coming from space. You would install an app on your computer and when you weren’t using it, its processing power was diverted to the app for computing purposes. Now, fast forward to 2025: could something similar be done e for running LLMs on personal computers? This could then generate usage credits applicable to the AI provider.
-
I am curious to explore other providers of generative artificial intelligence services. I am a paying subscriber to ChatGPT, but I haven’t seriously explored other alternatives yet, though I plan to do so. I fear missing things out if I ever leave OpenAI behind. I tested Perplexity, but I don’t believe in their business model. I think Claude from Anthropic seems to be the most interesting candidate. They have an iPhone and iPad app, just like OpenAI. From now on, I will systematically query both ChatGPT and Claude for my research needs and compare their performance.
-
I started documenting before-and-after comparisons of different betas of Apple Liquid Glass. Search for “beforeandafter” in the document to locate those comparisons. Honestly, things are improving. That wasn’t expected.
-
Discovered a fun Easter egg in iPadOS 26 beta6 Preview.app. It’s documented as exhibit #50 right here.
-
A Strange & Frustrating Mac App Store Issue
Experiencing issues re-downloading apps from the Mac App Store after testing them through TestFlight, leading to frozen downloads and frustration. Continue reading →
-
Am I starting to like Liquid Glass for the good and the bad things? Like they say: with beta6, it might be growing on me.
-
People are excited with generative AI and this is legitimate but this excitement overshadow a question we should ask more often about genAI (and about any new tech trends): what do we lose in return? My guess is that we lose a lot but we’re not sure what exactly. When we start to learn about the losses through experience and studies, we will question, hopefully, our use of AI in general. That’s my hope.
-
In my opinion, the best place for Liquid Glass is on the Apple Watch, because the overall graphic content is much more standardized and controlled. You rarely get weird backgrounds under UI elements.
