Amazon does work! It's actually one of our most used tools. We have a custom stealth stack consisting of things like in-house CAPTCHA solvers, cloud browsers, custom IPs, and adjustments to playwright that help us with matching normal web traffic.
Would recommend you try it out, seeing is believing
I'd been a software engineer for 25 years before language models came into the scene. There were so many parts of the engineering world I had no idea about: Large cloud data processing, lower level device / GPUs, UI frameworks and design.
I kept my to my niche.
Now I can build anything, but as I'm building it, I try to understand at least how it works. So I'm constantly asking questions. In the process, I think of my prior engineering skill as somewhat childlike. As I thought I was a decent engineer, I now see that I was decent at some things, but far from the kind of breadth that the wide engineering world has.
I think the takeaway is that you should always be curious and build things. Whether it's writing or discussing with an language model, writing software by hand, or creating software with a model and understanding it, I think you have the same effect.
> Now I can build anything, but as I'm building it, I try to understand at least how it works. So I'm constantly asking questions.
This is the way!
AI is the best technology ever invented for avoiding learning things. It’s also the best technology ever invented for learning things.
I think it’s very important to be on your guard for accidentally glazing over your understanding, but if understanding is your goal, there are so many more tools for this now.
Before AI, even when working collaboratively and bumping into concepts I did not fully grasp, there was so much more friction to get answers. Sure I could Google and try my luck, or I could ask a colleague. But the LLMs have infinite patience, can build diagrams, and can answer weird questions I would be afraid to ask.
Tencent Hy4 ties DeepSeek V4.1 Flash at 82/100, in 23.2 min for $0.30 vs 20.1 min for $2.26. Same game quality for about a seventh of the price. Cheap models keep closing the gap.
It used to be more interesting and elaborate, switching between different models. I used to use Cursor, then Codex, but since Opus 5.5 came out, I just use the Claude IDE. I got rid of all my skills and just set it to medium. Then I vaguely talk to it about very high-end things that I want, and it comes back with magic. Rinse and repeat.
I'd say the biggest trick is just to tell it what complete looks like. Be very specific about what you want it to come back with once it's done. That gives it a good idea what the goal is. It'll automatically create subagents and all that stuff. Although if you have money to burn and you're doing something large, you can tell it to just go wide or boil the ocean.
Also, make sure you're working on at least five different things. Otherwise, you're sitting around for one agent.
I can do 5 things concurrently for about a day max. 3 things concurrently all week and my weekends are spent untangling my brain. 3 things per day on Mondays and Tuesdays and 2 things per day concurrently on Wed-Fri and I feel ok but crave holidays.
I probably have some kind of pysccosis. I've had a personal $200 plan from one of the AI providers since March 2026. I'm never not making things now, I haven't rebooted my laptop for weeks because there is always an agent building something.
I cycle between 4-5 work projects and 2-3 personal projects. Every night I tell me agents I'll be going to bed and to not ask questions and keep going. I've always loved building things, but this feels more like an addiction than healthy sometimes.
They're not all done, and half the projects I bench, but this is where my personal projects live. https://wooten-labs.com
not gonna lie, you don't need to be doing these projects. they're not really worth it. instead of building an app to help you meet up with your friends somewhere central just go meet up with them and explore together.
I don't need to be doing all of these projects, but I enjoy doing all of these projects, much more than I enjoy going somewhere to hang out with people. Not everyone is the same.
Just to be clear, my comment about this being gross is about the rudeness of telling someone else their hobby is a waste of time, not about hanging out with friends.
> I haven't rebooted my laptop for weeks because there is always an agent building something.
I'm pretty pilled on cloud agents these days. If I've got 5 agents going at once, each is in their own dedicated, isolated cloud environment. There are a bazillion services for this now ampcode, hoplite.sh, cursor's cloud agents, devin.ai cloud, and more.
I haven't had an agent do anything from my laptop in a month or two.
> Every night I tell me agents I'll be going to bed and to not ask questions and keep going.
Most harnesses now support something like this:
/goal "Work on the current plan until 11 AM in my time zone, provide a summary of all the overnight work then."
Other than that, look after your own wellbeing! Nothing wrong with letting agents work overnight, just don't overextend yourself and don't spend all of your free time on it.
My friend and I have built portfolio management agents and pit them against each other to see who can make the most money (not real money just a simulated bank account and stock portfolio). So mine runs every morning to read the news and do analaysis and all that. I have it write a daily blog that I read to keep track of what it’s doing.
A daily blog for status and reasons for decision making works pretty well for tracking an agent that works unattended. It’s also kind of funny when it writes things like “well I got that one wrong” and “this is not what I expected to happen” hah
How long have they been running for and do any of them consistently beat index funds? Was always curious about the idea, but thought the market would already price it in.
They've been running for a few months, my friend's was beating SPY (VSP500) for a few days but not any more. Mine never beat SPY or VOO (Vanguard SP500 index). Now, let me tell you about my little sister. She says she just does what ChatGPT tells her to do and she's #1 on the leaderboard beating everyone including the SP500.
stewbot-v1 = my friend's bot
stonker = my bot
https://stonks.jettdigital.app is my little vibecoded stock market sim website, the bots work with via API. Each account starts with $10k and then you get a $500 deposited each month. If you want an account drop me an email ( see my profile ).
There are so many quants working 24x7 to eek out any advantage they can in the market, it's basically noise. If you want to algorithmically (or using AI) gain an advantage in market, pick a smaller one, like a niche betting market of some kind, that isn't overrun with other actors already doing it.
Yeah, I've been working on this thing for months. I'm trying to get it more polished before I release it. The reason the GitHub link is borked is because I need to have part of it be private, because some of the bot detection can't be open source, or it'll stop working.
They look like awesome apps*, as long as you're balancing your time by using them (they're outdoorsy and social!) I'd say you're doing fine. Take some extended time off if you need!
> I haven't rebooted my laptop for weeks because there is always an agent building something
I have a cheap mini PC I use as a devbox. Beyond meaning I can close my laptop and close all agent windows, it also keeps agents out of my personal shit.
respect. this is how i've been working the last 4 months or so. it's a bit crazy making and i do think i'm enjoying things less because i'm preoccupied, but i sure do feel productive. lol.
I have no idea why anybody is downvoting this. I honestly believe the OP is doing it right. Very few people get this level of productivity out of AI and the people who really "get it" with AI have this strong attraction to it. I estimate, fewer than 5% of programmers working with AI ever get here.
The problem is that I do want to interrogate the planning and implementation. Ask questions about trade offs. That means context switching. I don’t want my knowledge of the system to degrade rapidly.
I tried a T3 code for a hot second, but found it too buggy and missing features I missed from Claude Code Desktop. I do desperately need to figure out some way to get stuff off my laptop.
Yep, Claude & Codex desktop are innovating extremely rapidly. Some of the features like code review are hidden under commands, but basically obsolete whole heaps of AI code review startups when you use them.
I just ask Claude code to do a code review of PR X and it spawns a sub-agent. No hidden commands or anything special.
If I hint it might be security sensitive code, it bumps the effort level and burns my usage.
But it works quite well.
The findings are often debatable or unreachable states but it does find stuff and improves the overall quality I find.
What's funny though is that it almost always inevitably finds something wrong, even if you just ran it in another window and fixed everything it found. There is really a point of diminishing returns.
For anything non trivial I have Claude do a design doc with me. Then I tell it to “carefully review” the design which turns up edge cases. Then I tell it to make a step by step implementation plan document. Then ask it to review both again. Then I do a personal review and then off it goes on implementation while I pull up HN. It’s a lot like my team lead role at work. Btw, people who have a few years of being in a TL role are best positioned to make full use of the agents as well as teach others how to get the most out of them.
Claude Desktop's UI seems pretty similar to Cursor now, almost. I'm still on Cursor but if I use it daily I often hit usage limits before the end of the month. Would you say using Claude directly is better bang for buck?
I switched from Cursor to Claude Code (via T3.codes) because I was running out of usage on Grok 4.7. I regret not making the switch sooner, Opus 5.5 is much better than Grok, and Grok is decent!
Yeah, I agree, but Opus 5.5 burns through a ton of usage in Cursor so I rarely use it. Just wondering if cost wise it's better to use Claude Code than Cursor w/ Grok - sounds like you think it is?
Same journey for me, although I still use skills. Claude Code is fine; the remote-control feature is why I stick with it now. Generally anything in the cloud doesn't have a proper GPU that supports WebGPU, and as nice as SwiftShader is often you want to actually run things on a real machine. remote-control is by far the easiest way to do that from my phone.
> I got rid of all my skills and just set it to medium.
Yeah, exactly. I find increasingly little use for skills. I'd put it down to switching projects and stacks a lot but maybe... they just aren't as useful any more?
I wasn’t really using skills but then our project management tool added an MCP so I made a skill for grabbing tickets from the backlog, creating worktrees with the right naming convention, and moving ticket states for me automatically.
At the same time when I’m exploring new feature requests I now use the skill to write up the requirements and estimates into as many tickets as needed.
any prompt that you’re running a lot, make it a skill. i don’t make a lot of new ones. but still am using ones i made months ago and have been refining.
I use Claude Code Web almost exclusively for personal projects -- I’m guessing that’s pretty much the same interface but in the cloud -- and hardly ever run anything locally. Nicer UI than a terminal, real sandboxing, and it doesn’t tie up my computer. It’s just a remote agent that can make GitHub PRs.
But even worse for private stuff than using cloud inference to begin with, since it presumably has to clone the repo, have write & Actions access back, etc.
That's the bit that's common, I say it's worse because it has to actually check out the full source on their server. i.e. similar to if you used Claude locally, plus used third-party CI like Circle instead of Actions.
I modeled the UI explicitly after tiktok and youtube shorts on desktop.
For wikitok, I made it take up the entire screen, but for this project I felt that having the comments was a core part of the experience so it's better to have them both side by side to get context, so the mobile-like layout wins.
reply