prologic

Hey! It's at least 1 9's right? That's better than zero 9's LOL

For some reason I thought instacart as an e-commerce platform, not an online grocery store lol

Why? Wha does Doordash have to do with Instacart? How are these even remotely in the category of merging and buying the other?

It isn't just making indie software viable, it's creating a strong market (unsure if it's profitable) of solopreneurs that are now able to have a whole team of agents to design, build, and validate their ideas with very small budgets.

For example, I've been working on commodifying self-hosting for the past 10 years, and whilst there are numerous attempts of the same and healthy competition, there isn't quite anything that I would consider remotely close to what I invision. With Claude Code over the past year I've been able to 10x the progress and get to the point where hopefully by end of year (2026) I'll have my first batch run and realize my dream.

It's great!

My view on this is simple. Fundamentally, these "agent"(s) are effectively a function that computes the most probabilistic next token, or, what's the most likely next thing given the context and prompt.

Therefore, if you start prompting and adding to the context and session history a bunch of "useless" hyperbole, slur and profanity, unless the underlying language models have been trained on such things, I find it hard to reason about what it might predict the next series of tokens should be that would be remotely useful to you.

Ultimately, I really think you need to not treat it like anything, other than a powerful inferencing tool that has a great understanding of a very large corpus it has been trained on and has modelled very well. Use it to your advantage, there is no point in rudeness or profanity, it has no feelings, it isn't intelligent. At worst you risk garbage output.

What do you define as "AI Slop"?

It's a good question. If by "AI Slop" you mean, you ask ChatGPt to translate something from your native tongue to English so you can participate in a Reddit forum and the result is, probably correct, but a lot of "unwanted test you didn't ask for"?

What if it were you or me (I'm a native English speaker). Why if the tables were turned the other way around? I'm not even sure how I'd honestly be able to communicate with a person say from France, as I know a lick of French really, so I'd be reaching for some kind of translator, anything really. How do I avoid the wrath of those that would ben me over "AI Slop"?!

What's so bad about HTML anyway?

Marko looks like a mix of at least 3 languages. No thanks!

Don't you think it's pretty discriminatory to outright ban non-English speakers, even if said person(s) use AI to help translate their "ideas"?

I'm sorry, non-English speakers too?! da actual fuq?!

When you figure this out, please let me know!

Why do I need to pay for a plan for what appears to be a desktop app? Is the Taskboard a remote SaaS over the Internet that I configure my other Claude/Codex agents to to to and the Desktop App is, what, just a frontend to what's going on?

public repos don't have to be scraped or cralwed all over by bots right?