Vanclief

I support this pet theory, I tried out to reduce the output of Claude models with a "ADHD" prompt that made its responses small and to the point, but I could notice it degraded in performance as the session went on.

So I think what is going on is that because responses are part of the context window, those long/technical responses help it keep focus/attention.

We have been working hard on the past two months implementing agents for different tasks. We started with Claude code, as we really liked working with hooks. However being vendor locked and having usage limit problems, we ended up implementing our own "runtime" that keeps the conversation structure while having hooks. Right now it only works with OpenAI but its designed to be able to incorporate Claude / Gemini down the road

We ended up open sourcing that runtime if anyone is interested:

https://github.com/Vanclief/agent-composer

While I like the self reflection from this article, I don't think his methodology adds up (pun intended). First there are two main axis where LLMs can make you more productive: speed & code quality. I think everyone is obsessed about the first one, but its less relevant.

My personal hypothesis is that when using LLMs, you are only faster if you would be doing things like boilerplate code. For the rest, LLMs don't really make you faster but can make your code quality higher, which means better implementation and caching bugs earlier. I am a big fan of giving the diff of a commit to an LLM that has a file MCP so he can search for files in the repo and having it point any mistakes I have made.

Exactly, I had the same issue when I was younger. I thought if I read code I could not understand was because the other dev was a legend. Now its the opposite, I am amazed by code that does its job well, its understandable and has low complexity.

Claude Code is my favorite way to use LLMs for coding.

However I feel what we really need is to have an open source version of it where you can pass any model and also you can compare different models answers.

(Aider and other alternatives really doesn't feel as good to use as Claude Code)

I know this is not what anthropic would want to do as it removes their moat, but as a consumer I just want the best model and not be tied to an ecosystem. (Which I imagine is the largest fear of LLM model providers)

Stars are just a signal. When I am looking at multiple libraries that do the same, I am going to trust more a repo with 200 starts that one with 0. Its not perfect, but I don't have the time to go through the entire codebase and try it out. If the repo works for me I will star it to contribute to the signal.

Where did you get the data that there has never been a profitable one? Not calling you out, but curious of where you are getting this data.

I would expect that there have been multiple nuclear power plants that provide a net positive return, specially on countries like France where 70% of their energy is nuclear.

The comparison isn't even close. TikTok's relationship with the Chinese government is well-documented, not "supposed". They are legally required to share data under China's National Intelligence Law. The Chinese government has also a track record of pushing disinformation and find any way to destabilize Western democracies.

Douyin (The Chinese Tiktok version) limits users under 14 to 40 minutes per day and primarily serves educational content, while TikTok's algorithm outside China optimizes for maximum engagement regardless of content quality or user wellbeing.

US tech companies pursuing profit at the expense of user wellbeing is concerning and deserves its own topic. However, there is a fundamental difference between a profit driven company operating under US legal constraints and oversight, versus a platform forced to serve the strategic interests of a foreign government that keeps acting in bad faith.

I am all for preventing teen pregnancy, but the data sounds more like young people can't afford having children so they are waiting longer.

That used to be my flow, but I just added Avante to my Neovim and it has been a awesome for working with a single file.

If I need to do something more high level or that requires multiple files I still copy and paste to Claude / ChatGPT.

https://github.com/yetone/avante.nvim

Could you share with me the link of the profile you saw? I did work with Blockchain in the past but this current SaaS has nothing to do with that.

We recently applied to the current YC batch and got rejected. Seeing this just made me die a little inside.

We are a small SaaS that has very happy paying customers and a huge market. We solve a boring problem, with boring technologies and we are not the next OpenAI or Stripe. Yet we have easily a 10,000 X potential.

I feel like YC now prioritizes funding things that can be hyped more instead of actually funding things that can be solid software businesses.

If you are not a software engineer, you can't judge the correctness of any LLM answer on that topic, nor you know what are the right questions to ask.

From all my friends that are using LLMs, we software engineers are the ones that are taking the most advantage of it.

I am in no way fearful I am becoming irrelevant, on the opposite, I am actually very excited about these developments.

I strongly disagree with this statement, Golang is excellent for building backends and just the orchestration of all the services you are calling is better done in a typed/compiled language with great multi threading support and fast compilation times that is exactly what Golang excels at.

While from a first instance, this package seems a bit overkill, I think the idea is interesting and is something that can be improved for Go.

I also felt that Go errors where too bare-bones, so I developed a small package (https://github.com/Vanclief/ez) based on an awesome post that I saw here once. I use this package in all Golang code I touch.

Its hard to take this article seriously when it does not provide any real technical argument and has typos such as "egineers".

Fronted: React Backend: Golang Database: PostgreSQL

Why should namecheap subsidize the cost? That makes 0 sense and I say it as a namecheap customer. I don't see how its fair for them to cover the cost and I don't want them to do so because I am a happy customer and don't want them to go out of business or lower their product quality.

> The customers of those users absolutely do not want this. If I ever find out a company uses something like this, I'll no longer be doing business with them.

I second this 100%

This for me is a major red flag. It could maybe be mitigated if OP would require customers to run their own instance of cc.dev.

I don't think that its bad for the motivated engineer, its just a trade off between expressiveness and simplicity.

I personally find myself choosing go because I feel I am way more productive with the language. By keeping me constrained, I have less choices to make on how I solve the problem, and focus more on solving the problem.

Wait, children?

Unfortunate, but it was pretty obvious this would happen. The blackouts lasting only 48 hours was a mistake anyways. The only way I see them changing course is if enough people:

1. Delete the official app

2. Stop buying reddit awards, subscriptions, etc.

3. Mass delete their comments (Ideally after backing them up so that they can restore them later)

4. Stop visiting reddit all together

From reddit, this could easily be solved by offering "Reddit Pro", where you pay X per month, get perks and gives you an API key that you can use in any third app you like. Everyone is happy.

I strongly relate to this. I built my desktop computer and I run arch as my daily driver. For my phone: I just want a device that works out of the box. I don't want to think about it, or need to configure anything from it.

Thats the only reason I keep buying iPhones.

Are they trying to maximize their income in the next year or in their lifetime?

I would honestly advice them to change their mindset from trying to maximize income, to maximizing learning. So instead of finding a job that pays a lot, find a job where you learn a lot. Over the long term you will earn more if you are highly skilled.

1. I try to break the problem into smaller problems. 2. Pick the hardest one from those smaller problems and apply step 2 until it can't be broken into smaller pieces. 3. Once I have a "constrained" hard problem, I just try to solve it however I can.

If I solve that problem, that normally allows me to build from there, otherwise I may have to re-frame the problem. But by just starting, I get more information and ideas on how to tackle it.

At least the AI won't complain about the refactor haha

I completely agree that groundbreaking technologies come from an iterative process. However in the case of LLMs I believe we are already at a point where we can judge where the technology is going as its not the first iteration. Sure it will keep getting better and I think that its already a very useful tool.

My problem with it is that they are over hyping its capabilities and trying to market it as "it makes developers 55% faster" because it writes the code for them. I think it would be a better approach to market it as a great tool for automating repetitive tasks and a better way to consume documentation.

Exactly and I personally think that will always be the largest limiter to how good can the technology get. No matter how good the stochastic parrot gets, its still a parrot.

Could you provide an example of this? What kind of prompts are you using, and what does a complex system of different state-of-the art technology would look like?

Sincere question as I don't get how to use it for things larger than simple scripts.

> LLMs aren’t just the biggest change since social, mobile, or cloud–they’re the biggest thing since the World Wide Web. And on the coding front, they’re the biggest thing since IDEs and Stack Overflow, and may well eclipse them both.

I personally feel the technology is over-hyped. Sure, the ability of LLMs to generate "decent" code from a prompt is pretty impressive, but I don't think they are biger than Stack Overflow or IDEs.

So far my experience is that ChatGPT is great for generating code from languages I not proficient in or when I don't remember how to do something and I need a quick fix. So in a way it feels like a better "Google" but still I would rank it as inferior than Stack Overflow.

I am also hesitant about the statement that it makes us 5 times as productive because we only need to "check the code is good" for two main reasons:

1. It is my belief that if you are proficient enough in the task at hand, it is actually a distraction to be checking "someone else code" over just writing it yourself. When I wrote the code, I know it by heart and I know what it does (or is supposed to do). At least for me, having to be creating prompts and then reviewing the code that generates is slower and takes me out of the flow. It is also more exhausting than just writing the thing myself.

2. I am only able to check the correctness of the code, if am am proficient enough as a programmer (and possibly in the language as well). To become proficient I need to write a lot of code, but the more I use LLMs, the less repetitions I get in. So in a way it feels like LLMs are going to make you a "worse" programmer by doing the work for you.

Does anyone feel that way? Maybe I am wrong and the technology hasn't really clicked for me yet.