In this week’s newsletter, I wrote an essay about why real writers need not worry about Anthropic’s latest watermarking tool. You already embed your own “watermark” with your personal style, passion and soul. As for the weekly recap, I wrote about Reid Hoffman’s latest post on Questmaxxing, vibe coding with Gemini 3.7 Flash, training ChatGPT’s browser to comb LinkedIn content, building another Excel tool with Claude, geeking out with a boutique gaming handheld, and my eyerolling take on a McKinsey survey of HR people.
Weekly recap
Questmaxxing is a wonderful, wonderful article from Reid Hoffman. One of those things I wish I read when I was younger. The irony is that young people don’t like to read nowadays. Well, too bad. You can read the excerpts here or the original Substack post here.
The endless release of new AI models continues, this time with [Google’s Gemini 3.7 Flash](https://www.linkedin.com/posts/iantan1976_google-just-launched-gemini-37-flash-and-activity-7493840792214765568-UOZO. I vibe coded with it and was impressed by the sheer number of useful features it built without me asking.
I tried out the web browser within the ChatGPT desktop app and found that I had to train it like its kindergarten teacher. . I wanted it to trawl LinkedIn like a media monitoring agency and it’s getting there.
Another adventure in creating offline Excel tools with Claude.
The pain and pleasure of being a PC geek. Come see the OneXPlayer 3, a gaming handheld that is actually a powerful PC.
This survey about top future skills by McKinsey makes me roll my eyes.
Some lifehacks. Unfortunately, the first lifehack (a big discount on a Python training book) has expired.
Deep breathing, calm mind. Remember that, PSLE kiddos.
Sunday essay
In the past week, Anthropic announced that it would begin watermarking its text output due to EU regulations. I saw that some people online got upset by this, because they don’t want to be caught in a situation where their work may be lightly edited by AI but may now seem like it’s totally AI-generated.
I was nonplussed by the news. Firstly, these people are worried about the wrong things. Secondly, I know AI doesn’t use Gen X words like “nonplussed”. Thirdly… I’ll get to that at the end.
The real value of a written artifact is not who wrote it with what tool, but what is valuable in it, and how the content is clearly delivered to the audience.
A writer who knows how to write for his audience, and not just for himself, does not worry about whether he has used a pen, a typewriter, Google Search, Microsoft Word or ChatGPT in the writing process. The tools are simply what we use to transform our thoughts into words and our words must be able to transfix and transform you.
Most people who write (notice how I avoid calling them “writers”) do not understand what it means to write for the audience. If you have read this far, it is because you feel that each sentence is worth reading. Otherwise, you would have jumped off and I would have failed at my writing. Or maybe you are not my target audience and you have subscribed to the wrong newsletter.
No matter how good AI gets at writing, it cannot predict what I am going to write in the next few lines. Or how I will edit my text. This essay has been edited several times, and the previous version was more belligerent because I wrote it late in the evening after several days of poor sleep. But this version is more cordial because my brain has had some good rest. Our personalities are embedded in our words, and our words are also shaped by our current physiology and environment. The watermark that we embed in our writing is both consistent and dynamic.
If you want to be a writer, then write daily with your heart and soul, and distribute your content widely to find out if people will respond to your writing. Don’t worry about using Gen AI to clean up your typos, because Anthropic has also explained that it is unable to watermark writing that has been lightly edited by AI. It’s the people who rely too much on AI who should be worried. Do read Anthropic’s piece, it will teach you the nuances about how Gen AI works.
And if you do have a client who insists that you shouldn’t use AI, then write without AI. In this scenario, if your work does get flagged for AI usage even though you didn’t use it, then you must be writing like a robot. Then be human lah.