Thoughts on Tech News of Note - 02-06-2026

• Task-Oriented Agentic Programming
• Spotify Gets Bookish with Bookshop.org Partnership and Page Match
• TikTok Deemed Addictive by the EU

BTW, it seems like at least some of the posts on Moltbook (see last week's TTNN or pretty much anywhere on the internet if you've no idea what I'm talking about) were created by humans masquerading as OpenClaw bots. This is both disappointing and encouraging because we've no idea what posts are real and some of the posts are funny (those you want to be real) and some are terrifying (the ones you pray are fake). Just another item on the list of you-can't-trust-anything-you-see-on-the-internet.

Task-Oriented Agentic Programming
Using AI tools for programming is old hat now and many people, including those who have no idea how to write a single line of code, are using them to create all kinds of applications. But the process hadn't always been easy. Some people were using multiple AI tools to cross-check and improve upon the work of another tool just to get closer to the results they desired. And so, it turns out that getting the best results means not just better prompting but creating requirements that the AI tool can use to create the code to achieve the desired outcome. As an old programmer who spent many a day with a business partner to get an understanding of the pain points in their existing processes so the software I created would resolve those issues and automate repetitive and often tedious tasks, I know what this is. We used to ball it a "BRD" or "business requirements document". Oh man, the many days I spent writing, reviewing, editing, re-writing, re-reviewing, re-editing, and eventually cursing out (in my head, of course, because I'm a genteel lady who does not curse out loud) BRDs as a business analyst and then the even more days I spent as a young programmer trying to work through the obvious holes in those BRDs and having countless meetings to refine the approach so the final code would actually do what the business group thought they needed it to do. I say it that way because what they thought they needed it to do was almost never what it really needed to do.

Yeah, I am getting sidetracked here. The point is that to get better results from AI programming tools, you need to set better guardrails and be more specific about functionality and features. This means better planning up front by breaking that information down into a list of tasks that the AI tool can complete one by one, stepping through them until it has achieved the goal. One of the early tools to help automate this process is called Ralph. Ralph will take your PRD (product requirements document) wherein you define those features and functions, and it breaks them down into tasks for the AI programming tool. As it's described, it strikes me as much like Agile for AI. Ralph is creating and prioritizing the stories and the AI tool is iterating through them, ensuring it meets the prescribed conditions necessary for the story to be considered "done". And as the stories are completed, functionality is added to the program until you have your minimum viable product (MVP), and then on it goes from there because software is never finished. There are always bug reports, new feature requests, unexpected scope creep, etc. The key difference is you're the one pressing the AI tool for changes and you won't have to wait 6-12 months for them to be implemented.

But Ralph is already old hat itself. Updated versions of AI programming tools have built this functionality right in so that you don't have to rely on an external tool anymore. You can build your own little Agile AI team and iterate through stories as long as you like because the AI tools will never fight with you about the value and purpose of the latest new feature you want added or complain about not having time to attend the daily stand-up because there's too much work to do, or worse, attend the daily stand-up meeting and waste everyone else's time. The addition of this functionality allows for more complexity in the final product. It takes more upfront time in planning, but anyone who has ever worked in software development knows that more time well spent upfront translates to less time spent later on trying to fix things.

I've seen many articles and videos from pundits and experts saying that you'd still not use these tools to create massive applications to replace corporate America mainstays like Workday, ServiceNow, or Salesforce, but theoretically, it would seem on paper that we could get to that point. The question then becomes not so much whether the tools can make software that complex and valuable, but what is the total cost of ownership if you do it? The software companies that make these applications employ many engineers and designers to constantly fix bugs, add new features, and update the user interface as they learn from their users. At what point do you run into diminishing returns because you need humans to oversee certain aspects of the software lifecycle? AI tools aren't yet expert in user experience and interface design because they lack the human perspective. People want to use tools that are functional and also hopefully, beautiful in some sense. No one enjoys using an ugly app even if it functions perfectly. Where do you inject the human touch into this iterative AI programming approach? Or does all of this lead to a place where design barely matters because the only users interfacing with the software are other AI tools?

Now I'm laughing and crying to myself imagining someone sending an OpenClaw bot out to file a service request in ServiceNow. Good luck little bot.

Spotify Gets Bookish with Bookshop.org Partnership and Page Match
I am an occasional and reluctant Spotify user. For a while, I was a paying subscriber, and I made an intentional and determined effort to use it for music and podcasts. I'd left Tidal because it didn't support podcasts and trying to make what I considered the best of it with USB digital audio converters that supported MQA (Master Quality Authenticated, the Tidal high-quality format) became more annoying than enjoyable. I thought I could consolidate my music and podcast player into one app and be happy. I was wrong. Although Spotify works fine for podcasts, I never warmed up to it. I preferred Podcast Republic and Pocket Casts, even though subscribing to premium podcasts was sometimes easier to do in Spotify. And once I discovered Qobuz for high-resolution music with its access to the modern equivalent of CD liners, I was done paying for Spotify and I was back to using dedicated apps for specific functions. Spotify tempted me back briefly when it added audiobooks and I did indeed try to listen to a few. But I'd always remember how many audiobooks I'd already bought and not yet listened to in Audible and I eventually once again stopped using Spotify. Nevertheless, I do really like that Spotify has audiobook access, so the latest announcements about their partnership with Bookshop.org to allow you to purchase a copy of the physical book to whose audiobook you are listening and the ability to sync up physical and digital books with audiobooks is almost exciting. I say almost because I'm still not tempted to go back to Spotify to try it. But the idea of syncing a hardcover book through the magic of OCR and AI is genuinely surprising and delightful to me. I buy most of my books in digital form nowadays, but every once in a while, I come across a book I want to own in the true sense of ownership. I want to put it on my bookshelf or lay it on my living room table. Or maybe I want to share it with someone else (and pray I get it back someday). As I ponder it, it's probably a hassle to keep a hardcover book and an audiobook in sync in reality, but the idea of it is lovely and I hope Spotify keeps adding new features and pushing other platforms to innovate in new ways. This feels to me like something Amazon Music/Audible/Kindle should have been doing years ago, seeing as they sell both physical and digital books.

TikTok Deemed Addictive by the EU
Once again, a story pops up on my radar well after I thought I'd nailed down the list for the week. But the articles I read about this were just too much for me to ignore. It seems that the European Commission has preliminarily deemed that TikTok is intentionally designed to be addictive and needs to change to avoid fines.

Stop laughing. It's a real story.

The Digital Services Act (DSA) is just over three years old. TikTok is just over nine years old. So this issue would appear to pre-date the DSA itself, yet this is something on which they've just now landed. The features that the EU called out - infinite scroll, autoplay, push notifications, and its recommendation algorithm - are not new features. Even more, they are the core features of the app. What the EU seems to have an issue with is the entire point of the app. TikTok is meant to be sticky and draw you in with more of what you like to watch. Indeed, most social media apps work this way. If TikTok were to disable or remove these features, it wouldn't really be TikTok anymore. One might as well use something like Skylight Social or Loops.video. And that is not meant to be a dig against those tools. Those tools are purpose built to be alternatives to TikTok and Instagram that are free of some of the concerns the EU is raising. And so they are underdogs for a reason and by intent. People want to use TikTok because it's low friction. Switching over to another app, especially one that doesn't have the same type of recommendation algorithm or the same popular creators and influencers, would be viewed as a downgrade to the user experience by most. It will take time for alternatives to TikTok to gain ground unless rulings like this force them to be less competitive and/or more nations continue to ban TikTok.

But the thing that really bothered me about this story is that despite how I and others may feel about TikTok, the company built an app that is used and loved by many. Those people use it and love it because of the very features that the EU is pushing against. It seems very baby-and-bath-water to just say what has made TikTok great just has to stop. It seems that there should be some middle ground here. After all, if grown folks (I'll leave children to the side for a moment) want to watch TikTok for hours on end and not be plagued with enforced breaks or stymied by being recommended videos they don't want to see, why should the EU just be able to stop them? We let grown folks drink and smoke despite the risks to their health. In many places, gambling is legal despite the risk to people's bank accounts. I can hear in my head many of those grown folks swearing at their computer screens, wondering how dare the EU tell them they can't have their TikTok time. I'm not sure at the moment what the best approach is here. It seems easy to say that there should be an age limit and children below that age limit should be subject to screen time limits and be less encumbered by the ability and temptation to scroll endlessly, but what is the right remedy for adults whose brains are supposedly fully developed and are making their own informed decisions? The EU suggests that certain vulnerable groups of adults should also be protected but I see no way to implement that protection without additional infringements of privacy and decency. What is the test to prove you're not a vulnerable adult?

I guess I now have to say as they do on news sites that this is a developing story because this is definitely not the end of it.