The simplest metric you should be tracking is; has it generated income.

In this sense my use of agentic coding has performed very well. People asking for evidence and repos are a little naive to how capitalism works in the real world. If I’m making money on something I’m not going to let you copy it, and I’m sure as hell not going to devalue it by publicising that it was built by AI.

It would be madness to work completely offline these days, and all of these systems have version history and document recovery built in.

If the person can use AI to lead a noticeably better life, something that may have been impossible previously due to economic circumstance, then the first order benefits outweigh the second order drawbacks.

I’m not disputing what you’re saying, I just think that treating it like a zero sum game every time the conversation comes up is showing an immense amount of privilege.

You, me, the parent commenter; we’re all dying, we don’t have time to optimise for the best outcome.

I think the article itself serves as enough evidence to prove you should ignore their advice.

I had assumed it was just simple mathematics and the belief that cash is the easiest way to transfer charitable effort. If I can readily earn 50USD/hour, rather than doing a volunteering job that I could pay 25USD/hour to do, I simply do my job and pay for 2 people to volunteer.

If Deepseek is free it undermines the value of LLMs, so the value of these US companies is mainly speculation/FOMO over AGI.

I spend a lot of time doing what I would describe as an AI VLOOKUP where I get an embedding for every record in Table A and lookup the top matches from Table B, sometimes adding in some LLM synthesising of results. The problem I have is there is always a manual step because the results are never perfect, so I might get AI to classify 10k rows of data and then I’ll go through and sort them and start replacing answers I think are incorrect etc.

It seems to be quite an obvious little niche but I think part of the problem is cost. A lot of the tools which are monthly subscriptions can’t afford the cost of the requests, so they resort to simple labelling. I think some kind of BYO Model might be the way.

Anyone seen or working on such a thing?

Comments URL: https://news.ycombinator.com/item?id=45519311

Points: 1

# Comments: 1

Geohot is easily a 99.999 percentile developer, and yet he can’t seem to reconcile that the other 99.999 percent are doing something much more basic than he can ever comprehend.

It’s some kind of expert paradox, if everyone was as smart and capable as the experts, then they wouldn’t be experts.

I have come across many developers that behave like the AI. Can’t explain codebases they’ve built, can’t maintain consistency.

It’s like a aerospace engineer not believing that the person that designs the toys in an Kinder egg doesn’t know how fluid sims work.

Kids learn to speak before they learn to think about what they're saying. A 2/3 year old can start regurgitating sentences and forming new ones which sound an awful lot like real speech, but it seems like it's often just the child trying to fit in, they don't really understand what they're saying.

I used to joke my kids talking was sometimes just like typing a word on my phone and then just hitting the next predictive word that shows up. Since then it's evolved in a way that seems similar to LLMs.

The actually process of thought seems slightly divorced from the ability to pattern match words, but the patter matching serves as a way to communicate it. I think we need a thinking machine to spit out vectors that the LLM can convert into language. So I don't think they are a dead end, I think they are just missing the other half of the puzzle.

When I'm in the terminal I can call on Agents who can create standardised documents so there is a memory of the product management side of things that extends beyond the context window of Claude.

It guides you through the specification process so that you have extremely tight tasks for Claude to churn through, with any context, documentation and acceptance criteria.

Perhaps there are others similar, but I have found it completely transformative.

For my money it's by far the best Claude Code compliment.

For those struggling with impairments it must be hard to continue to adapt to your phone shape shifting with each update.

External links are bad for user retention/addiction.

This also has a side effect of back linking no longer being a measure of a 'good' website, so good quality content from inconsistently trafficked sites gets buried on search results.

I think once the sheen of Microsoft Copilot and the like wear off and people realise LLMs are really good at creating deterministic tools but not very good at being one, not only will the volume of LLM usage decline, but the urgency will too.

In the same way Claude Code is a different beast to Cursor, my own process is a different beast to Claude Code and the months I've spent building out a robust pipeline is now paying dividends.

I also think someone at The Register needs to go on a statistics course. Those figures seem to paint the picture that an overwhelming majority of those surveyed have had positive outcomes, which I don't think is represented by the slightly snarky headline.

- Makerkit Next.js/Supabase Starter Kit

- Python backend processing

- BMAD framework for building specification

- Claude Code with Max subscription

- Cursor for in IDE adjustments

I have managed to make some pretty incredible tools, it definitely feels like magic.

I would say I split my time 70% using BMAD as an assistant to build out my scope and clarify what I am trying to do in my own head, then 30% supervising Claude Code.

I have also managed to build out more simple tools using Streamlit to great effect

It would seem on first glance that you can't set up a social network called Friendster just because you bought Friendster.com domain.