DrSiemer

My own tool has a Project Starter, that takes users from idea to todo list in a number of very specific phases. I've been using it for small to medium sized projects for a year now and it still works great for me.

Just don't expect to end up with a finished project; it's more like a first draft. Once it's there, it's much easier to determine what it is you actually want, since you can directly experience what works and what should be changed.

One important caveat is that I do not work with agents; each step goes through a fairly rigid manual review phase.

Yeah, not a great experience, but it will let you play if you just register with something random.

That only strengthens the example. Their recommendations are usually awful.

I'd argue even that is done better on underground services. Plus they'll just show you actually useful metrics, like the IMDB rating.

Europe is full of people who are not fans, but will still hear all about the games when their country plays, because it's basically impossible to avoid. Chat groups, work banter, waiting rooms, everybody will be talking about it ALL THE TIME. There is a base level of knowledge for the sport everybody shares. Maybe it's the same with American Football in the US? As a European, I still have no idea what that is even remotely about.

A game having two main uninterrupted halves is one of those foundational facts. Players train for being able to play for that long without getting too thirsty to continue.

Being able to distill a story down into bite sized chunks sounds like an optimization for the younger generation, with their 20 second attention span.

It's not something I necessarily associate with "good writing". Or does the skill transcend the medium and help you write better long form content as well?

Haha, yep, first thing I thought of as well. I still have a offline copy of that exact site saved somewhere.

One of my side projects is a tool that lets you control your entire system with STT. It's built on Whisper and supports hot swapping custom profiles, so you can add easy commands for any software.

I intend to use it to work on low stakes vibe coding projects while I'm doing other stuff. Todays LLMs are a lot better at interpreting rambling dictation with mid-message corrections.

There are a few paid programs out there that do the same, but they made my vibe slop sense tingle and are not aimed at development.

The one advantage of actually trying to use SO was that the fear of asking a question usually made me do so much research that I'd solve my issue in the process of fully describing it.

That did also make the community lose out on the answer though.

It would be pretty cool if it could convert a splat to a super compact high performance 3D ASCII widget

That money is for the research, not salary

Wealth, not income. The statistic is about net worth growth. Saying "that's wealth, not income" doesn't refute it. It just clarifies the mechanism.

If the system is rigged heavily against you, relying on it to affect change does feel like a losing strategy.

The fact that such a large part of the population supports literal murder could also be considered a political statement. One that would not have been expressed so strongly without what happened.

So much of this madness could be resolved with a simple income cap. Musk’s wealth grew by $1 million per minute over the past year. Who can seriously argue that this is fair and balanced?

CodeMerger: https://codemerger.nl I've never liked the lack of control I feel using agents or tools like Cursor or Antigravity. I found myself having much better results simply pasting full source code in free chat, so I built a tool around that philosophy and I've been exclusively developing with it for a year now. It includes a Project Starter and code architecture analysis tool.

Presentable: a photo library sorter with Ai powered organization assistance, a compare canvas with various viewing modes and customizable folder sorting shortcut templates (wip).

AlwaysWhisper: a tool that let's me attach STT to my entire OS, with custom wrappers for different programs, adding theoretical voice control to any software (wip).

ScreenLoader: an Electron based tool that can load any web source as a kiosk app, full of useful features like keep-alive, covering multiple screens and tracking output logs.

Inputboard: a unified all-inputs hardware board, that transfers input data to any prototype I want to work on using an optocoupler, so I won't have to fiddle around with setting up clean and reliable inputs from cheap Chinese components every time I just want to test something.

Squire: an agentic board game helper, that can ingest a manual and will hopefully help decrease the time spent on endless discussions about seemingly conflicting rules. It should also be able to help me play a game when I don't fully understand the rules myself yet (wip).

NodeRunner: an agent that plays the WikiGame, focusing on speed, efficiency and token usage (the result of a fun competition with a colleague).

Sonic Bloom: more of an experiment than a tool. It's a wireless piece of custom hardware, that listens to conversations, sends data to an LLM through fast STT and returns a color choice that matches the topic being discussed to the hardware, which then controls an LED ring. It also has a small display that explains the logic behind the color choice.

Image-to-story: a VLM tool that kickstarts a written story using an image and has some rudimentary tools to expand on it based on user instructions (wip).

At-work-or-not: and Android app/website where colleagues can check if I'm working from home, if I'll be at the office or if I'm not working at all. Also doubles as a private record for tracking transport expenses.

SharedMaps: a Maps based website where groups of people can share custom categories of geo locations and drop comments on them.

VMG: an image format that includes audio with images and offers TTS input to easily add narration.

Who wants Coffee: a small Android app to help me remember who wants to drink what when I go for a round at the office.

And a pile of Python scripts for smaller useful tasks.

A random experiment I made a few weeks ago does something similar, but it only uses the real cursor position for now: https://2shine.nl/demo/mousemaze/

Plenty of annoyance in here for sure. Looks like 17 cannot be finished on mobile though. Switching to desktop view resets progress.

Adding a deadline to a disclosure of a vulnerability of this nature is standard practice. Every day it's not patched is a day data could be compromised. Any halfway competent lawyer should be fully aware of this.

Disclosure without a deadline WILL be ignored.

It does not matter if it's Google or your local boyscouts club, any organization requiring users to provide information that can be abused in the wrong hands takes on a responsibility to handle such data responsibly.

That sounds cumbersome and even more wasteful than my own method of simply dumping a fixed selection of project code in Gemini for each set of requests. Is there any benefit to pruning?

Recent models have started to "fix" HTML issues with ugly hacks like !important. The result looks like it works, but the tech debt is considerable.

Still, it's just a temporary hindrance. Nothing a decent system prompt can't take care of until the models evolve.

This is exactly how I feel about it. The cognitive load of starting a new project is so small now. It's also made it very easy to switch between projects, something that took way too much headspace to do on a whim in the before times.

We'll get to the point where we can mass moderate core knowledge eventually. We may need to hand out extra weight for verified experts and some kind of most-votes-win type logic (perhaps even comments?), but live training data updates will be a massive evolution for language models.

So "Corporate profit taxes are a game of hide-and-seek", but "you can tax the value flow, the revenue generated in-country, or the massive energy consumption of the data centers"?

The author acts as if taxes are not a completely fluid system, that will quickly adept to ensure revenue keeps flowing, squeezing wherever the squeezing is the juiciest. It does not need cautious calibration.

The way I work with LLMs is a bit different.

I use a custom tool, that basically merges all my code into a single prompt. Most of my projects are relatively small, usually maxing out at 200k tokens, so I can just dump the whole thing into Gemini Pro for every feature set I am working on. It's a more manual way of working, but it ensures full control over the code changes.

For new projects I usually just copy the llm.md from the tool itself and strip out the custom part. I might add creating it as a feature of the tool in the future.

A few days ago I tried to use AntiGravity (on default settings) and that was an awful experience. Slow, pondering, continuously making dumb mistakes, only responding to feedback with code and it took at least 3 hours (and a lot of hand holding) to end up on a broken version of what I wanted.

I gave up, tried again using my own tool and was done in half an hour. Not sure if it will work as well for other people, but it definitely does for me.

ChatGPTs implementation of Memory is terrible. It quickly fills up with useless garbage and sometimes even plain incorrect statements, that are usually only relevant to one obscure conversation I had with it months ago.

A local, project specific llm.md is absolutely something I require though. Without that, language models kept on "fixing" random things in my code that it considered to be incorrect, despite comments on those lines literally telling it to NOT CHANGE THIS LINE OR THIS COMMENT.

My llm.md is structured like this:

- Instructions for the LLM on how to use it

- Examples of a bad and a good note

- LLM editable notes on quirks in the project

It helps a lot with making an LLM understand when things are unusual for a reason.

Besides that file, I wrap every prompt in a project specific intro and outro. I use these to take care of common undesirable LLM behavior, like removing my comments.

I also tell it to use a specific format on its own comments, so I can make it automatically clean those up on the next pass, which takes care of most of the aftercare.

The tech works well enough to function as an excuse for massive layoffs. When all that is over, companies can start hiring again. Probably with a preference for employees that can demonstrate affinity with the new tools.

Most of the people that dislike genAI would have the exact same opinion if all the training data was paid for in full (whatever a fair price would be for what is essentially just reference material)

Your desired setup is just a single prompt away...

If it's really hidden in its own special tab and NEVER comes out unwanted, I can see some benefits.

It's just not a very good fit for Firefox. I assume it would run on a cloud service, which is very much a privacy issue. Especially because it appears to be something "free", making my data the product.

Bots in the Hall. Neural Viz. The Meat Dept video for Igorrr's ADHD. More will come.

You need talented people to make good stuff, but at this time most of them still fear the new tools.

It's not the most energy efficient workflow, but I work on relatively small codebases and I made a tool that let's me dump all of it in an LLM with a single copy/paste. This works surprisingly well with Gemini 2.5 Pro (1.000.000 ctx).

The only real mistakes it makes are some model specific quirks, like occasionally stripping out certain array index operators. Other than that, it works fine with 150.000 token size conversations. I've gone up to 500.000 with no real issues besides a bit of a slowdown. It's also great for log analysis, which I have maximized to 900.000 tokens.