Pedro Martins

1.2K posts

Pedro Martins

Pedro Martins

@nikuscs

🇵🇹 Doing stuff. Typescript, React, AI . Building : https://t.co/DILygdVkQj, https://t.co/HzLd4WBtMX, https://t.co/DZo2ZCRQUn, https://t.co/czk5c7iPs0

United States Entrou em Eylül 2009
2.9K Seguindo10.4K Seguidores
Tweet fixado
Pedro Martins
Pedro Martins@nikuscs·
Introducing Bookmarkjar ✨. I had thousands of saves and could never find “that one link.” A calm place to save, and quickly fin your stuff again. bookmarkjar.com More info: 👇
Pedro Martins tweet media
English
11
9
28
5.1K
Rita 🇵🇹🇻🇦
Rita 🇵🇹🇻🇦@southevropa·
My feed is now filled w people from all over the world getting traumatised by Brazilians They finally get why we don’t like them lol
Rita 🇵🇹🇻🇦 tweet media
English
63
6
169
18.3K
Evan You
Evan You@youyuxi·
That’s basically vanilla WoW. And it was actually fun when you get sent on a quest to the other side of the world without flying mounts! Still remember vividly the journey I took with my friends to go from Eastern Kingdoms to Kalimdor, running and battling through PvP territories, taking the boat, landing in a completely different looking area… such good times. Never had that experience again in later MMOs.
English
5
0
86
5.6K
Punyapal Shah | Laravel | PHP | Dev | OpenSource
Nowadays when I open my for you feed... - Which database you use - Which AI Agent you use - Claude or codex - Which email provider is best - Supabase is best database - Which is best hosting - Mac or Linux - Vibe coding or agentic coding - Interviewer if cache is faster why we cannot cache everything... - ChatGPT or gemini - GitHub is down - As a developer what are you thinking about this and that... - You have $100/20 what you will choose... And bla bla bla I cannot find anything useful yet.
English
4
0
13
1K
Pedro Martins
Pedro Martins@nikuscs·
@tomfgoodwin Right, its just insane. Some of the posts dont even have linebreaks, just large chunk of AI slop text with hashtags at the end lol
English
0
0
0
109
Tom Goodwin
Tom Goodwin@tomfgoodwin·
I would pay so much money to sit down with the head of LinkedIn and just show them my feed, and ask them to explain what I see An infinite numher of monkeys on infinite typewriters, would produce largely more valuable content.
English
141
54
1.7K
43.6K
Pedro Martins
Pedro Martins@nikuscs·
this. This and “🚨Breaking: “. And also tons using the daily bait posts that still get a ton of likes and idkn why. Like “I have 20$ to spend codex or claude code”, “just got a new macbook what i should install first” . I mute a ton of those daily and somehow they still endup in my timeline lol
English
0
0
3
1.4K
Pedro Martins
Pedro Martins@nikuscs·
Im getting pretty exhausted about all this AI companies doing shady stuff, they keep cutting corners left and right in order to reduce costs. I totally get that @AnthropicAI got a massive boom, and need to reduce the usages and so on, thats totally acceptable, but can we please get a "stable" model/usage-limits for more than 3 months please? Im at the point where im gonna throw all this shit back and go back to the basics until we get a decent AI company that keeps their words. Timeline: - Opus 4.6 Released - Damn good! - Massive Marketing - OMGLUL we getting to much users, who have though? 😂 - We need to reduce usage - Everyone mad, cut OpenClaw, all good people got along. - Massive nerf to thinking or whatever they did, vibes don't lie. We back to sloopy model acting like GPT-3. - OH wait, the solution is to use Opus for "brain" and "Haiku" reducing as much as 20 cents, great, here we go again, workarounds for cutting costs. I'm giving @OpenAI codex a fair try starting next month and hopefully OpenAI doesnt fall into the same mistakes as Anthropic did, otherwise we might as well start looking into Chinese Model & Opensource more and more, because in the long term, companies cant rely on flaky models every 15 days.
English
2
0
8
2.3K
Pedro Martins
Pedro Martins@nikuscs·
@NoemiTitarenco @AnthropicAI Yeah a few runs of compose 2 per day will quickly burn into 50-70$ monthly budget quite easy. But i totally get it, at least stuff there doesnt look nerfed :p
English
1
0
0
10
Pedro Martins
Pedro Martins@nikuscs·
Right, iv been saying this. Not only this, but also it seems like companies are laying off thinking AI will do most of the work. But while you output 10x faster, you also need more senior devs to actually review this work, otherwise "we" devs cant keep with some much context of wahts being done in short time. The bottleneck is there :p
English
0
0
0
398
Klaas
Klaas@forgebitz·
having mid developers is now going to slow your company down before a mid developer would do mid tasks like crud monkey work, forms, etc, which is fine, that needed to be done now you have a mid developer being able to generate massive amounts of slop code, but they don't know it's slop, so the better/senior devs need to review the slop creating a bottleneck of vibe coded slop
English
100
31
802
101.6K
Vitalii Dodonov
Vitalii Dodonov@vitaliidodonov·
Dear algo, Please show this tweet to founders, builders and creators in Europe.
English
126
4
369
19K
Pedro Martins
Pedro Martins@nikuscs·
Pedro Martins@nikuscs

Im getting pretty exhausted about all this AI companies doing shady stuff, they keep cutting corners left and right in order to reduce costs. I totally get that @AnthropicAI got a massive boom, and need to reduce the usages and so on, thats totally acceptable, but can we please get a "stable" model/usage-limits for more than 3 months please? Im at the point where im gonna throw all this shit back and go back to the basics until we get a decent AI company that keeps their words. Timeline: - Opus 4.6 Released - Damn good! - Massive Marketing - OMGLUL we getting to much users, who have though? 😂 - We need to reduce usage - Everyone mad, cut OpenClaw, all good people got along. - Massive nerf to thinking or whatever they did, vibes don't lie. We back to sloopy model acting like GPT-3. - OH wait, the solution is to use Opus for "brain" and "Haiku" reducing as much as 20 cents, great, here we go again, workarounds for cutting costs. I'm giving @OpenAI codex a fair try starting next month and hopefully OpenAI doesnt fall into the same mistakes as Anthropic did, otherwise we might as well start looking into Chinese Model & Opensource more and more, because in the long term, companies cant rely on flaky models every 15 days.

English
0
0
1
1.5K
Evan You
Evan You@youyuxi·
Honestly don't know what happened to Claude Code. Tried a one-off simple task on a fresh directory yesterday, tried a bunch of things that didn't work, asked for a ton of permissions, and then got stuck for 4 minutes before I got tired of waiting and killed the session. This was on medium effort. Switched to Codex gpt 5.4 with medium effort and one shotted the task in under 1 minute.
English
237
95
2.8K
327.9K
Pedro Martins
Pedro Martins@nikuscs·
I don't let AI take into the foundation of the code or have strong opinions most of the time in serious projects. Mostly using to explore ideias and dog fooding boring parts of implementation tbh. Iv been using Codex mostly just to review / devils advocate till today. Opus / Anthropic vibes were really nice, until they are not. Cursor tbh is still goat, specially composer 1/2 is a really great model for "dumb" stuff & quick analysis, but it burns out pretty fast for me. Gonna force myself to use Codex more and see where we get.
English
1
0
1
43
Noemi
Noemi@NoemiTitarenco·
@nikuscs @AnthropicAI FWIW - I still use Cursor Auto the most, and I have both CC and Codex unlimited plans. GPT-5.4 is really smart, but I like working through problems with Cursor better. Like I literally just enjoy using my brain more. 😂
English
1
0
1
48
Pedro Martins
Pedro Martins@nikuscs·
@kenwheeler loool, my exactly tough this morning. If this shit cant be "stable" i rather get this going back to the basics again lol
English
0
0
1
92
patagucci perf papi
patagucci perf papi@kenwheeler·
these models are so nerfed i’m about to write this shit by hand like a psychopath
English
32
6
322
6.7K
Wise
Wise@trikcode·
Anthropic's new strategy is smart use Opus to think use Haiku to execute 11% cheaper and actually performs better
English
133
34
1.9K
62.7K
DHH
DHH@dhh·
In celebration of Omarchy 3.5 being the first distro to ship with complete Linux compatibility for the new XPS Panther Lake laptops, @Dell made me a special unit with super + omarchy keys instead of Windows and Copilot. So damn cool!
DHH tweet mediaDHH tweet mediaDHH tweet media
English
185
236
6.7K
323.3K
Pedro Martins
Pedro Martins@nikuscs·
Sending prompts ( basic ), see what terminals are open for a worktress/space ( Claude code /remote-control kindly does this already ), but more important, able to review a bit what code is being produced. While vibe coders wont "look" at the code, its often at least for me see what is the current state of changes to correct or align on the fixes i want.
English
0
0
0
307
Theo - t3.gg
Theo - t3.gg@theo·
I want to control my agents from my phone. I'm sure you do too. What workflows do you want? Do you have one that works now? Tell me all about it.
English
531
19
1.7K
238.4K
Elon Musk
Elon Musk@elonmusk·
Tesla driving itself around LA
English
12.4K
31.2K
376.7K
118.1M
Pedro Martins
Pedro Martins@nikuscs·
@theo Mine is "lets explore first, no code" dont think "Thoughts" didnt worked for me after a few refinements :D
English
0
0
0
450
Theo - t3.gg
Theo - t3.gg@theo·
It’s cool seeing my heroes getting so deep into practical coding with AI
Theo - t3.gg tweet media
English
16
5
395
30K