Well, 2 very intensive and educational months on every level later, here’s another summary:

• the mentioned tools have proven themselves, additionally I’ve added opencode as an agent and quite a few MCP servers.
• the quality differences between models are enormous. Claude models are in a league of their own (also price-wise). Through opencode I’ve gained a lot of experience with Kimi and Minimax (available for free for one context window there), which deliver quite usable results but can also go spectacularly off the rails. Due to lack of funds I’m currently mainly using the very affordable GLM4.7, which for example really struggles with Lisp-like code. I’d even go so far as to propose the thesis that LLMs have certain language preferences. Claude tends towards Python, Minimax towards bash, etc. I wouldn’t let the openrouter/free models near any relevant code.
• MCP servers (and other plugins/tools) significantly change the vibe coding experience. To avoid having the same servers running everywhere, I can recommend an mcp-proxy.

From a developer’s perspective, LLMs are extremely powerful tools that will massively change this profession. I’m more than happy to be able to experience this exciting time, just like being there at the beginnings of the internet in the early 90s. The agent-based approach has led to a paradigm shift. I suspect it will now be about composing these agents, linking them together, etc. There are many approaches to this, one simple idea is to integrate sub-agents via ACP. Goose is currently taking this path, Hermes or Pi have been able to do this for longer, AFAIK.

A week ago, the CEO of a company I’ve recently started working for gave me a Claude Code license. Since then I’ve been riding a learning wave like I haven’t experienced in a long time. The best combination of tools I’ve found so far consists of: OpenRouter, Goose, and agent-shell. (Guix packages) Very powerful tools that have already saved me an enormous amount of time and effort. Currently I’m experimenting with MCP servers like emacs-mcp-server and others, but I’m still not sure whether and how useful that is…

Sixtus has struck again! I always enjoy seeing his stuff; big fan.

Probably the best documentary on AI I’ve seen recently.

Edit 20.3.2026:
In addition to the dystopia, here’s a worth-watching positivist perspective:

From the popular series Loneliness Kills:

A woman once told me that she occasionally feels so lonely that she pretends to be asleep on the train so that the conductor will touch her on the shoulder.

Since there are repeatedly misunderstandings about distinguishing empathy from compassion:

Have I mentioned before how great I find it when ancient insights are scientifically confirmed? ;)

PS: Here is also a blog post by Matthieu Ricard on this topic.

Truth is a complicated matter. That’s why caution is warranted when someone comes along with simple “truths” and claims them for themselves. Basically, this only works in the natural and technical sciences (see wiki).
At least with everything biological, I’m a strong proponent of Radical Constructivism and especially with humans, I think it’s inhumane to deny them their own truth, i.e., to apply a strictly technical/mathematical definition of truth:
Let’s assume (e.g., in a war or something), a person sees something for a few seconds that traumatizes them for the rest of their life. Other people see this too, but it doesn’t affect them. For the former, this brief moment (over time) becomes “truth” because it changes their life. For the others, truth might be the sum of individual moments, which physically of course they’re right about. But to then tell the traumatized person: “You didn’t experience truth, because physically much more happened. Your truth is false/incomplete etc.” I personally find inhumane. If this is additionally argued with great self-evidence and conviction, one has to ask oneself whether and how much empathy is still present…

You can truly be glad about humanity’s development when you consider that speciesism has not only been overcome but even surpassed, because there are apparently people (presumably mostly female, PETA activists and tendencies misanthropic) who no longer deny animals or certain species their feelings, but humans: For them, the feelings of animals (especially horses) are automatically authentic, to be taken seriously and treated sensitively. Human feelings, on the other hand, are to be questioned, analyzed and tested for rationality and value. Speciesism 2.0 or Post-Speciesism?

Very reassuring that not only I find the current mood (psychologically and sociologically) dangerous, but that professionals (here: a psychologist) apparently see it this way too (or can see it).
At one point he even asks “have we learned nothing from 1933?” For such a comparison, I’ve been tarred and feathered in all chats and conversations so far (which for me is again a clear indication of how emotional and feeling-driven any discussion about this has become…)

Mir macht die derzeitige Polarisierung und das Suchen nach Schuldigen Angst:
Ich verstehe zwar die emotional und menschlich nachvollziehbare Reaktion von Wut, Enttaeuschung, Hilflosigkeit, Resignation und vielem mehr, aber deshalb eine nicht unerheblich grosse Bevoelkerungsgruppe auf ein einziges Kriterium zu reduzieren um dort eine imaginaere Schuldfrage zu klaeren erinnert mich an dunkle Kapitel der Menschheit. Wir sollten gelernt und begriffen haben das so etwas (soziologisch und psychologisch) sehr, sehr gefaehrlich ist und vor allem zu ueberhaupt nichts Gutem fuehrt, ganz im Gegenteil…

“…Eigentlich sollte es eine Binsenwahrheit sein, dass eine Gesellschaft nur gemeinsam und miteinander durch eine solche Krise kommen kann. Wer (auf welcher Seite auch immer) zu einem Kampf “Wir gegen Sie” aufruft, sägt an dem Ast, auf dem alle gemeinsam sitzen und übersieht einmal mehr die Wissenschaft…”