I keep seeing signs that using AI is becoming more and more expensive: reduced tokens budget for the same pricing might be a trend now.

As someone who’s pondering whether I can and should buy a Mac to develop native apps on it with Claude Code, this trend makes me pause a little bit. Hardware is now very expensive and if you add more pricing pressure on AI subscription this adds up.

One question that comes to my mind is: what capabilities a 20$ a month subscription of Claude buys me when developing native iPhone apps?

As we are getting ready for our trip, my wife and I has been using Siri AI more than usual for things that, frankly, I wasn’t sure would work. But 90% of the time, Siri AI accomplished the task. Example: fetching the Aeroplan number and flight reservations numbers to fill a flight registration form in Safari. We couldn’t do that before with the older Siri. It’s encouraging.

Now that almost anyone can build a native iPhone app using vibe coding or more sophisticated AI-agent methods, companies that keep releasing shitty apps for their products should pay attention. Synology is one example. They make great NAS devices¹, but their iPhone apps aren’t great. Unlike Ubiquiti, they don’t offer an iPhone app for managing their devices. I came across this app on Reddit this morning, and it looks well done. That guy essentially replicated Synology DSM capabilities in his app. That’s cool and useful. If Synology cannot hire good iPhone developers, fine, but with AI-based coding, they have no excuses to try a different approach.

I barely knew you. Fully removed all traces of this cancer from my M4 Mac mini. I feed much better now.

Lately, because token usage in subscriptions has become costlier, I’ve become more selective about the models I choose for prompts. I switch between models like Haiku, Sonnet, and Opus based on the complexity of the prompt and the effort needed to produce a high-quality response. Is anyone else adopting this approach?

Another set of thoughts about OpenAI’s recent announcements, which trigger all sorts of reactions in me.

• OpenAI seems to be trying to be more like Apple; it’s obvious in their slides during their keynote.
• I think OpenAI shows courage by doing live demos of things that are clearly not ready for prime time. Since people would like Apple to return to live events, I’m not sure I want to see those kinds of failures, as they might overshadow the message.
• I saw a different Sam Altam during the keynote, who appeared more enthusiastic than usual, but it felt a bit off.
• Many of the new features are available only to Pro plans, which I’m not willing to subscribe to because they are too expensive.
• They make progress on their desktop app, but it still feels bloated, just like Anthropic’s desktop client. None of them are native apps, yet they brag about AI coding capabilities and still depend on Electron. Go figure.
• You might find Dots a cute name for agents, but the last thing I want from AI is to be cute. I would argue that I want an invisible AI.
• I still don’t see a use case for agents like Dots or Muse (Meta). I don’t trust that technology to take over my digital life, even for the simplest task. I don’t feel good about asking an AI bot to check off an item on my to-do list. And I certainly don’t want an AI bot to do shopping for me. BTW, who’s responsible for buying errors: the bot, the AI company or the one paying for it?

I haven’t felt this anti-AI attitude as strongly as I do this week. I wonder why.

I enjoy writing in my personal life and at work. In the latter case, it’s an important part of my responsibilities. The IT department is currently deploying and assigning Copilot Premium licenses. I’m still waiting for my turn, but from what my colleagues tell me, the experience of using Copilot with access to our documents, emails, conversations and notes is quite positive and almost magical in some situations. I look forward to trying out Copilot, especially since I write much more than most of my colleagues¹ and since writing is their main fuel, I’m up for a few positive surprises. I have a feeling that I will finally benefit from my consistent efforts.