• I made two important changes this morning for my automation-related operating environment:

    1. Using DigitalOcean monitoring, I created two resources alerts (CPU > 50% for 5 min, Disk usage > 70% for 5 min). Alerts will make me look into n8n automation misbehaviours.
    2. I switched my AI nodes to use Claude AI Haiku 4.5 instead of Sonnet 4.5 to reduce costs for comparable results. I don’t think my summarization tasks needs more powerful LLM.
  • When Things Go Wrong With AI-Generated Code

    My first bad experience: the code generated by Claude Code made my dashboard unresponsive in my browser. Eventually, the data stopped updating. After a ten-minute debugging session, I asked Claude Code to revert the change, and it did so promptly. But then I started getting execution failure notices on Discord. A lot of notifications. Then I started investigating… It appears the browser was making frequent refresh requests to one of my workflows, which depleted my Claude pay-per-use credits. Continue reading โ†’

  • Ten days into 2026, I have achieved much more than I anticipated. If I maintain this pace, I will complete my list of wild ideas soon. It’s not just about checking items off the list, but also about learning a lot along the way. It’s very fulfilling.

  • I’m making good progress on my personal dashboard idea. While people may criticize LLMs and enjoy coding for fun, for me, it opens up possibilities I couldn’t have imagined before.

  • One of the frustrating aspects of LLMs is their lack of consistency unless you develop specific skills, which can take time to implement effectively. For example, I wanted to generate documentation for my most recent n8n automation workflow, but Claude was unable to do it, and I can’t remember the prompt that finally made it possible. I should have saved it somewhere for easy retrieval. I’m wasting precious credits. ๐Ÿคฆ๐Ÿปโ€โ™‚๏ธ

  • Simon Willison on AI-assisted programming:

    The more time I spend on AI-assisted programming the less afraid I am for my job, because it turns out building software - especially at the rate it’s now possible to build - still requires enormous skill, experience and depth of understanding.

  • My Defaults as of 2026-01-10

    Changes from the last edition are in bold. โœ‰๏ธ Mail Client: Fastmail ๐Ÿ“จ Mail Server: Fastmail ๐Ÿ“ Notes: Craft + Apple Notes โœ… To-Do: Things 3 ๐Ÿ“ท iPhone Photo Shooting: Camera.app ๐Ÿ“š Photo Management: Photos.app + Photomator ๐Ÿ—“๏ธ Calendar: Calendar.app ๐Ÿ—„๏ธ Cloud file storage: iCloud ๐Ÿ“ฐ RSS: Reeder connected to Inoreader ๐Ÿ“‡ Contacts: Contacts ๐Ÿ•ธ๏ธ Browser: Mobile Safari + ARC Browser on Mac + ChatGPT Atlas ๐Ÿง  AI: ChatGPT + Claude AI ๐Ÿ”Ž Search: Kagi Search ๐Ÿ’ฌ Chat: iMessage (WhatsApp when abroad) ๐Ÿ”– Bookmarks: AnyBox ๐Ÿ‘“ Read It Later: Inoreader ๐Ÿ“œ Word Processing: Ulysses, Craft ๐Ÿ“Š Spreadsheets: Numbers ๐Ÿ› Presentations: Keynote ๐Ÿ›’ Shopping Lists: Reminders ๐Ÿง‘โ€๐Ÿณ Meal Planning: None ๐Ÿ’ฐ Budgeting & Personal Finance: Numbers ๐Ÿ—ž๏ธ News: La Presse (Apple News for English news) ๐ŸŽถ Music: Apple Music ๐ŸŽง Podcasts: Apple Podcasts ๐Ÿ” Password Management: iCloud Keychain & Apple Passwords ๐Ÿ‘จ๐Ÿปโ€๐Ÿ’ป Blog hosting: Ghost, Micro. Continue reading โ†’

  • In today’s world, with LLM, Claude Code, etc., is Apple’s Swift Playground still relevant, even for younger aspiring coders? A few years ago, it seems we were hearing much more about it than today.

  • I’m building a Dashboard. ๐Ÿ˜Ž

  • Apparently, people are barely using Stack Overflow to ask questions, thanks to LLMs and AI. I expect a similar trend among people in a community like this one on Micro.blog. Some questions would be super easy to answer by asking ChatGPT or the like. I do understand that many people still want this human touch, though.

  • Something is about to happen. Again. ๐Ÿค—๐Ÿซฃ

  • Updated my n8n instance from v2.0.3 to v2.2.4. Super easy to do (I’m using the Docker Compose installation provided by the DigitalOcean 1-click install droplet. Took a droplet snapshot before, just in case something goes wrong. So far, so good. Of course, Claude helped me out on this. I’m not a Linux or Docker expert. ๐Ÿ˜…

  • On Tahoe Icons

    Just finished reading “Itโ€™s hard to justify Tahoe icons”, which many UI design pundits and non-UI experts, as well as simple, passionate Mac users, have been referring to a lot recently. I appreciate the documentation effort. It’s really well done. Very convincing. But… Designers age and are gradually replaced by a younger generation. Whether you like it or not, they bring new beliefs (justified or not), design principles (better or not), and values (questionable or not). Continue reading โ†’

  • I can confidently say I’ve learned at least one significant new concept every day over the past two to three weeks. It is not only satisfying but also quite stimulating. This will keep me from aging, that’s for sure.

  • Claude Code skills are probably the most intriguing aspect of Claude Code and Claude AI. I’m not so sure yet how to take advantage of them. My understanding from this excellent video is that you have to be an expert at something to create those skills.md file.

  • An Important Lesson

    When I started my studies in computer science over 40 years ago, we learned to read functional specifications and then translate them into machine instructions (COBOL, FORTRAN, Pascal, etc.). It was the training of a programmer. I knew that one day I could become the person who writes functional specifications. I didn’t become a programmer, nor did I work in the development world. Due to my recent experience with Claude AI, Claude Code, and Vercel to create custom applications, I realize that I have become the one who writes functional specifications, but for processing by artificial intelligence. Continue reading โ†’

  • Added a new and much-needed feature to my micro.blog front end. See my prompt below.

  • On OpenRouter.ai

    I just finished reading about the service openrouter.ai. I was curious to understand the purpose of this service as well as its business model. I saw several instances of this service being used in n8n workflows. The problem I see with this service is that it makes the consistency of the quality of responses from the requested LLMs even more unpredictable. Each request could be handled by LLMs with different characteristics and performance from one time to another. Continue reading โ†’

  • Another update to my blog: the blog post categories are no longer listed on the main page header. Categories are shown on each blog post. I think that’s enough. Categories are available in the Archive page. This makes the home page less busy.

  • My latest n8n workflow automates summarizing my Micro.blog timeline (via its private RSS feed) and sends the results to my Discord server every hour. Pretty cool, right? Auto-generated description: A flowchart illustrates a process involving an RSS feed trigger, RSS reading, post aggregation, interaction with an AI agent using the Anthropic Chat Model, and communication with Discord.