> "A typo or two also helps to show it’s not AI (one of the biggest issues right now)."

However, nowadays it is used as a synonym for everything that is somehow generated by an LLM. Regardless of whether it is a spec-driven, carefully reviewed and iterative piece of software or some yolo-style one-prompter with no idea how it was done.

Most of the time, I end up putting in more work than I get out of it. Onboarding, reviewing, and mentoring all take significant time.

Even with the best students we had, paying around 400 euros a month, I would not say that I saved five hours a week.

And even when they reach the point of being truly productive, they are usually already finished with their studies. If we then hire them full-time, they cost significantly more.

Code style is a Pareto-optimal problem space: what one person finds readable may look like complete chaos to someone else. There’s no objective truth, and that’s why I believe that in a project involving multiple people, spending time on this is largely a waste of time.

> […]

> Or it'll help me debug my code and point out things I've missed.

I made both of these statements myself and later wondered why I had never connected them.

In the beginning, I used AI a lot to help me debug my own code, mostly through ChatGPT.

Later, I started using an AI agent that generated code, but it often didn’t work perfectly. I spent a lot of time trying to steer the AI to improve the output. Sometimes it worked, but other times it was just frustrating and felt like a waste of time.

At some point, I combined these two approaches: I cleared the context, told the AI that there was some code that wasn’t working as expected, and asked it to perform a root cause analysis, starting by trying to reproduce the issue. I was very surprised by how much better the agent became at finding and eventually fixing problems when I framed the task from this different perspective.

Now, I have commands in Claude Code for this and other due diligence tasks, and it’s been a long time since I last felt like I was wasting my time.

I think this sums it up well. Working with LLMs is already confusing and unpredictable. Adding a convoluted RAG pipeline (unless it is truly necessary because of context size limitations) only makes things worse compared to simply emulating what we would normally do.

As I see it, this prompt is essentially an "executable script". In your view, should all prompts be analyzed and possibly blocked based on heuristics that flag malicious intent? Should we also prevent the LLM from simply writing an equivalent script in a programming language, even if it is never executed? How is this different from requiring all programming languages (at least from big companies with big engineering teams) to include such security checks before code is compiled?

Paul Erdös was told about this problem with multiple explanations and just rejected the answer. He could not believe it until they ran a simulation.

Recently, I realized that this applies not only to the first 70–80% of a project but sometimes also to the final 70-80%.

I couldn’t make progress with Claude on a major refactoring from scratch, so I started implementing it myself. Once I had shaped the idea clearly enough but in a very early state, I handed it back to Claude to finish and it worked flawlessly, down to the last CHANGELOG entry, without any further input from me.

I saw this as a form of extensive guardrails or prompting-by-example.

https://en.m.wikipedia.org/wiki/Mett

When I searched for the safe temperature for pork (in German), I found this as the first link (Kagi search engine)

> Ideally, pork should taste pink, with a core temperature between 58 and 59 degrees Celsius. You can determine the exact temperature using a meat thermometer. Is that not a health concern? Not anymore, as nutrition expert Dagmar von Cramm confirms: > “Trichinae inspection in Germany is so strict — even for wild boars — that there is no longer any danger.”

https://www.stern.de/genuss/essen/warum-sie-schweinefleisch-...

Stern is a major magazine in Germany.

But then I remind myself: it’s not a product made for me. I don’t have to like it. Clearly, the target group loves it. My kids have adored it for years. Even now, with my oldest having access to Spotify Kids, she still prefers her Toniebox in the evening before bed. The figurines aren’t just a medium, they’re toys in their own right. They’re shared, traded, and loved. And they really enjoy squeezing those silly ears.

Many other families in my circle tell the same story. Some tried similar products that launched soon after the original, often ones using cards (though not Yoto). But after a few weeks, their kids lost interest and asked for a Toniebox instead. (It reminds me of when my parents bought me a Sega Master System, even though all I wanted was a Super Nintendo.)