mattm

I agree. I speak from experience because I actually went down this path for about a month. It is tempting. However, what I realized after some reflection is that there are just so many decisions that need to be made along the course of developing a product that it's never going to be possible to just one-shot it. There's a continuous feedback loop that is needed and agents aren't going to be the ones filling that role.

@tox9cc - FYI you're flagged as dead. Not sure why, your past comments all seem valuable to the discussions.

In this last week I've started reviewing code more and in just a short time, I've found 3 fairly simple things that were introduced by Claude that were not wrong per se, however, they were very inefficient and didn't address the root of the issue. I still think we're in a place where the output looks good as long as you don't look under the hood or keep it scoped to small, vibe-coded projects. Once you get beyond that it can fall apart. Anthropic must have a large codebase by now though. Yet I haven't seen much released from them about how they actually work day to day on development.

> sometimes I don’t know how I want to write something until I’m finished

This is knowledge that goes back to the beginning of software development - "Plan to throw [version] one away".

I think this could potentially become a good practice. LLMs make it so easy and cheap to just get it working and build that v1. Then you can play around with it and see if works and read the code about what could be better. Throw away the LLM generated version and now this is the part where human expertise comes in. Based on what you've learned from the v1, now guide the LLM more closely about how to write the thing and help guide it so that making small PRs that are easily reviewable and understandable are the output.

Hey just some feedback. I'm giving it a try on a new project I'm spinning up. It does seem faster although I haven't timed it against Claude or anything. I like the integrated browser to immediately see the changes (or see what the agent is doing). Thanks for having Linux support.

I've been following Demis a lot over the past year and he's alluded to this in interviews that basically he'd rather be doing research than leading a team. I know a lot of people have tagged him for being the next CEO of Google but he doesn't really want that. This role seems more aligned with what he actually wants to be doing.

I think this is true more often than many people want to admit. One of the most famous examples being Lennon and McCartney. Even though they couldn't stand each other near the end that tension helped produce their best music. They never did anything comparable as solo musicians.

I've had a personal health dashboard tracking my metrics for the past four years. Yes I am quantifiably healthier than I was.

Yeah this is what I did. All my personal services are running on one EC2 instance and I use subdomains to separate them. If I need more memory it's easy to scale up. I just use sqlite on the instance for my database. You don't need a postgres service if it's just for you.

Hosted services are only great for setup and scaling. You're not going to need scaling. Infra setup can now be done easily with AI and terraform.

I'm a fan of Boox although I have the larger model.

I wrote a bit about this here as well. Don't really have the best solution yet but hope it helps

https://mattmccormick.ca/we-re-all-managers-now-my-journey-i...

Closer to 10 months actually - Oct (Sept if you include the pre-season) through June.

But just to add on for comparison, the NHL has 1312 regular season games. Plus 4 rounds of best-of-7 playoff series with 16 teams making the playoffs. Let's say an average of 5.75 games per round, that's 86 playoff games bringing us to ~1400 games per year.

Over 4 years, that's 5600 games. The FIFA World Cup had 104 games this year. It's actually pretty impressive that the WC can even make 40% while hosting < 2% of the games that the NHL has. Or put another way, 1 WC game is equivalent to 54 NHL games - 2/3 of one team's full regular season.

I know this isn't the point of your post, but as an aside, how come Intuit is an example of being grandfathered in? I wonder if we'll see the rise of plain text accounting[1] in an AI world as it can be easier for AI agents to deal with.

[1] https://plaintextaccounting.org/

Here's the full version https://www.eff.org/wp/digital-privacy-us-border-2017

One of the names in the byline matches his HN username so it's likely this.

Note: The HN headline says "US citizen charged" but the article's headline and content do not mention his status at all.

I recall a similar quote from Elton John that I'll paraphrase:

"I've had a lot of hits so you'd think I'd know in advance which ones will become hits. Songs that I was sure would become hits went nowhere and some songs that I didn't think anything of became my biggest hits"

It's been a long time since I heard this so I'm probably mangling it. A quick search shows that he probably did say something like this though https://www.birminghammail.co.uk/news/showbiz-tv/sir-elton-j...

The lesson is - just put it out there and see what happens

> I built a SaaS in a few hours. It does its job better than tools people pay real money for.

It's easy to build something for yourself that solves your own use case. You know how you work and what's important to you. Most of the effort in software goes into understand people other than you and building guardrails for them.

> Whatever niche tool you’re sketching this week, a thousand people can prompt a working version by Sunday night

A lot of people on HN are naturally builders. For them it's fun to play around with AI and see what can be built. Not everyone is like that.

For individuals, how many people actually want to spend their weekends building a niche tool for themselves vs just paying a few dollars a month to access something that does the job even though it may not be perfect?

For businesses, the buy vs. build tradeoff does now have some new compelling evidence in favour of "buy" but there's still maintenance needed for the service internally, there will be people with feature requests for it. Before you know it, the company is diverting resources away from their core business for a support tool that doesn't really drive impact.

I think we'll see more options in the marketplace as individuals will be able to get started more easily but SaaS is far from dead.

I also really like Boox. I'm on my 2nd Boox product now having upgraded from my old Max Lumi so that I could get colours.

Reading and understanding documentation and producing clear, written requirements will be more critical than it is today. We've always known documentation is important but it's frequently glossed over in many projects and not taken as seriously as code reviews. With AI though, it's going to be much more important that all requirements are written down clearly and captured.

Same here. I tried something similar to Superpowers and it went completely overboard for a small bug fix - writing a TDD, generating artifacts, etc.

Even the CEO's "apology" is pretty bad. He still finds a way to take shots at the original poster saying his original message was inflammatory (could also be read as how I'm justified in my response), that "he started it" and that the team was "spoken down to or treated dismissively" which they weren't. All the original feedback was about the project and was not directed at specific individuals.

Wow, you sound exactly like me - except I have a bit less years of experience. Let me know if you want to connect to help keep stay motivated (email in profile)

Your AI agent asks their AI agent

Bezos' parents lent him $250k to start Amazon. The point is that by the time Bezos started Amazon they were wealthy and could provide him this safety net. Not many middle class families would be able to loan their kid that much money.

Reminds me of a blog post I once read from a manager writing about all the qualities of being a good manager. I read it nodding along that they all seemed like good traits. Then in the comment section there was a post from someone saying something like "You were my manager at one point and honestly you were one of the worst managers I've had in my career. I didn't see many of these behaviours from you". The author responded with something like "I don't disagree. There's sometimes a gap between knowing and doing"

The same principle holds for quality management. You don't need to inspect every single product. However, if you inspect a small number of products at random, you'll detect a large percentage of the quality issues.

While leaders can't know everyone they should make it a priority to have those random connections outside their inner circle. If they don't, they become in danger of hearing only the info that their inner circle wants them to hear.

The whole Github page feels AI generated too. Working 120 hours per week for 9 months? Give me a break.

If murder was legal, surely the amount of murders would increase.

> Sycophancy, in which chatbots agree with and excessively praise users, is a trait they’ve manifested partly because their training involves human beings rating their responses. “Users tend to like the models telling them that they’re great and so it’s quite easy to go too far in that direction”

We've already seen the effects of social media algorithms that route people into isolated bubbles. This is going to exaggerate it.

The founder raised $8M on Wefunder and seems too have taken it and run. He hasn't provided any updates for a few years now.