Vlad Temian

@bdstanga totally agree, especially with the things that llms need human intervention imo, systems where humans still have a say are the ones that will pass the test of time

@vtemian IMO what changed is that the bar for a saas to work is a lot higher now. Simple software/tools no longer add value. Now the value added is mostly in non-software aspects like compliance, data, security, reliability etc, things that llms cannot easily do yet

i built 20+ developer tools in 6 months. all tested, documented, published. near-zero usage. not because they're bad. because i was competing with the user's own Claude session. blog.vtemian.com/post/disposabl…

@tomachristian my optimism is trying to convince me that we'll have better tooling for handling security and code quality in the near future but is just my optimism. in reality, we're bringing more complexity that is hard to handle and verify at such a huge scale we're approching

@vtemian You see what’s going on, especially in the JS ecosystem… the party you get the thing from doesn’t even have to me malicious on their own, it’s enough for them to lack care in their dependencies; but even with proper care….

@tomachristian curios about a thing: even if you would know, 100% sure, that a library was not written with any agent, would you still use it? new library, not an existing one. lack of trust might be the case for pre-AI gen as well

@vtemian The reason I personally avoid anything “out there” ATM, more than before, is lack of trust; and honestly I don’t know what would make me get over that. That is also one more reason for myself to build my own tools now that it take minutes

@abarbatei maybe is better to share the prompt or your claude session, than to open-source the tool *

and in your example with the cool tool locally, i argue that not the tool itself is useful, but the fact that you solved the problem in a certain way same with agentprobe. it is cool and fun that you can have a library that parses JSONL transcripts and transforms them into realtime streameble events but the fact that you know that, you can point to agentprobe's github repo and tell claude to do the same, or extract just the specific implementation details you are interested in is more about discovery than problem solving the fact that you can show other people that a problem can be solved in a specific way and they can use claude to do the same maybe is better to share the prompt or your claude session, the OS the tool

but the value isn't gone. it shifted. from dev tools (things humans use) to agent-ready tools (things AI agents call on your behalf). from building the tool to building the environment: hosting, integrations, previews for disposable tools.

claude-notes: converts Claude Code transcripts to HTML. but you just paste your JSONL into Claude and say "make this readable." 5 min. generic loses to specific when specific is just as cheap.

@martin_valchev_ openable.dev

Slow scroll, strong projects. Share what you are building so we can support and test 👇

@TriCast_ openable.dev

Hey builders! It’s the weekend 🌟 Drop what you’re working on below - whether it's SaaS, AI agents, OpenClaw builds, marketing, whatever you’re shipping Let’s connect 👇🏻

@KaiXCreator openable.dev

What are you building this weekend? Drop your project URL Let’s drive some traffic

@heyblake openable.dev

Drop your project URL Let’s drive some traffic

@ThePrimeagen claudebin.com

Hey, you got a cool project that you are building? Link it I want to yap about cool projects

@thdxr if you can't hit in-house, outsource it

the hardest part about having a remote team is you can't hit your employees

@AlexPalcuie @bunjavascript i just built bunless, an adblocker plugin for bun

@bunjavascript c'mon guys, I got three unskippable ads in a row while trying to build my slop AI agent

In the next version of Bun Bun goes ad-supported!

@zawasp fo shure

@vtemian Did you prompt it to talk like Macron? 🤣

might be the biggest unlock in months

@badlogicgames performative coding

at this point i'm labeling this as performance art

@cipriancaba not sure yet, but i hope it compounds quickly

@vtemian Worth spending time with them?