In fact, I’d argue the physical risk of loss, theft, or data compromise is much higher at home than in a professional datacenter with power redundancy, security controls, and constant uptime monitoring.

It’s a bit like saying, "Don’t trust the bank, they could take your money and freeze your account — keep all your money under the mattress." Technically possible, yes. But come on.

To me, self hosted also means I rent a machine with Hetzner and run the server software on it. Its cheap, stable, fast, secure and Hetzner wont screw me over with my data. I have a LOT less headache and I can rent a vserver for a long time until the hardware cost for a server running at home is surpassed.

I can also very simply assign a domain to it and am pretty sure that software like nextcloud offers oauth access so my friends would NOT be required to sign up for my "weird app". Well, technically they do but oauth automates it.

Am I missing something?

i've tried agent-style workflows in copilot and windsurf (on claude 3.5 and 4), and honestly, they often just get stuck or build themselves into a corner. they don’t seem to reason across structure or long-term architecture in any meaningful way. it might look helpful at first, but what comes out tends to be fragile and usually something i’d refactor immediately.

sure, the model writes fast – but that speed doesn't translate into actual productivity for me unless it’s something dead simple. and if i’m spending a lot of time generating boilerplate, i usually take that as a design smell, not a task i want to automate harder.

so i’m honestly wondering: is cc max really that much better? are those productivity claims based on something fundamentally different? or is it more about tool enthusiasm + selective wins?

You can't make up a couple of conversation topics and expect the LLMs to do the rest by just switching languages. People approach the same topics completely different in different languages. The app looks like someone picked a couple of topics and the rest is "just" ChatGPT advanced voice mode.

And the worst thing is that the LLMs in TTS do not sound native and cannot teach you pronounciation and learning to listen and understand (which is the whole point in having spoken conversation).

And the other way around, the STT will not notice pronounciation mistakes made by the student - so the app cannot tell you: oh, its pronounced like this.

This is also a huge problem for people having a terrarium with geckos or saurians - they see and need much different wavelengths than we humans do.

I am no expert for carnivorous plants - maybe they are fine but seeing that there is no UV emitting part in the lighting setup, there may be an important part of the spectrum missing for the plants.

I would recommend looking around on HuggingFace altough I found it a bit intimidating at the beginning. The place is just HUGE and they assume some knowledge.

I would also recommend creating a platform user on OpenAI and/or Anthropic and look up their docs. The accounts there are free but if you put a few dollars in there, you can actually make requests against their APIs which is the most simple way of playing around with LLMs imho.

Here are some topics you could do some research about:

- Foundation models (e.g., GPT, BERT, T5)

- Transformer architecture

- Natural Language Processing (NLP) basics

- Prompt engineering

- Fine-tuning and transfer learning

- Ethical considerations in AI

- AI safety and alignment

- Large Language Models (LLMs)

- Generative models for images (e.g., DALL-E, Stable Diffusion)

- AI frameworks and libraries (e.g., TensorFlow, PyTorch, Hugging Face)

- AI APIs and integration (also frameworks to build with AI like LangChain/ LangGraph)

- Vector databases and embeddings

- RAG

- Reinforcement Learning from Human Feedback (RLHF)

- AI model evaluation and benchmarking

- AI-assisted coding and code generation

Key points:

- Mental modeling is faster than actual coding

- "Mental coding" can happen anywhere, anytime

- Tools like Obsidian help organize thoughts before coding

- LLMs can be used as brainstorming partners

- Always keep the MVP in mind, then iterate

I'm curious to hear how other developers approach the planning phase of their projects. Do you have any unique pre-coding rituals or tools you find particularly useful? How do you balance planning with the desire to start coding?

Comments URL: https://news.ycombinator.com/item?id=41630565

Points: 2

# Comments: 1

Heck, it would even be against the law in my country and it favors companies over people...

We tried gather.town in my team (we develop web services for a big news site) and dropped it after a week or two. While some people seemed to like it, others disliked it very strongly. I personally disliked it as well:

It drew lots of resources of my PC. It was like having MS Teams constantly running in meeting mode. For me it felt childish to push a pixel avatar around in a virtual office. And also tiresome (its one more thing you have to keep in mind alongside your work). Having the camera on all the time is a no-go. And walking into the private space of a desk and talking to a person working there is as invasive and interuptive as it was before in the office. The app brought the worst of the office days into my remote days.

So all in all, it was a cute little game for a few minutes, but it made work feel miserable.