Moreno Antunes

@alistaiir Does this use a terminal per process? Running out of pty devices already on my day to day workflow

We're launching the Monitor tool in today's Claude Code release. Claude spawns a background process and each stdout line streams into the conversation, without blocking the thread. e.g. "Use the monitor tool and `kubectl logs -f | grep ..` to listen for errors, make a pr to fix any crashes" This is both more reliable and token efficient than polling within the agent loop.

A few weeks ago we gave every employee @Traba_Work — ops, sales, recruiting, finance — access to @claudeai Code connected to our dbs, design files, knowledge bases, sales pipelines, etc. 100+ people. Zero engineers. They built more in weeks than had shipped the previous year:

@alex_prompter Token burner

Meta found that forcing an llm to show its work, step by step, with evidence for every claim, nearly halves its error rate when verifying code patches the technique is embarrassingly simple: a structured template the model has to fill in before it's allowed to say "yes" or "no" no fine-tuning. no new architecture. just a checklist that won't let the model skip steps

@milan_milanovic This post is a few years outdated

𝗪𝗵𝗮𝘁'𝘀 𝘁𝗵𝗲 𝗺𝗮𝗶𝗻 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗸𝗶𝗹𝗹𝗲𝗿 𝗳𝗼𝗿 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿𝘀? Context switching. You send a colleague a "quick" Slack message. Takes you 𝟱 𝘀𝗲𝗰𝗼𝗻𝗱𝘀, but it costs them 𝟮𝟯 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 to get back into deep focus. Now imagine this happening 5-10 times a day. Your best engineers aren't shipping slow. We keep breaking their flow. It takes about 𝟭𝟱 𝗺𝗶𝗻𝘂𝘁𝗲𝘀 of uninterrupted work to reach flow state. One notification can break it, even if the developer never opens it. Here is what helps: - 𝗗𝗲𝗳𝗮𝘂𝗹𝘁 𝘁𝗼 𝗮𝘀𝘆𝗻𝗰. Not everything needs an instant reply - 𝗕𝗹𝗼𝗰𝗸 𝗳𝗼𝗰𝘂𝘀 𝘁𝗶𝗺𝗲. Protect 2-4 hours for deep work daily - 𝗕𝗮𝘁𝗰𝗵 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻𝘀. Combine five scattered pings into one message - 𝗗𝗼𝗰𝘂𝗺𝗲𝗻𝘁 𝗮𝗻𝘀𝘄𝗲𝗿𝘀. If it's asked twice, write it down once One team I worked with blocked focus time for a month. Story completion went up 𝟯𝟱%. Bug reports dropped 𝟮𝟴%. Stop treating your team's focus like it's free. It's the most expensive resource you have.

@Bender_MkII @paraschopra Well being #47 meant there were at least #46 learning opportunities before it vs none

"Virality is a function of attempts" is survivorship bias in a productivity hoodie. Thousands of builders shipped 50 projects and got nothing — they're just not writing threads about it. The attempts mattered because this one was good, not because attempt #47 was statistically overdue.

Ever wondered why OpenClaw went viral but many other similar projects didn’t? Well, just look at the number of projects by OpenClaw’s creator. Virality is a function of number of attempts. It’s so rare and unpredictable that your best bet is to maximize taking shots at it. Same is true with tweets/videos. You’d see that the fastest growing accounts are those that produce a ton, and not those that keep perfecting a single thing that they hope to go viral.

@levie The interface and experience will be fluid and contextual, fully adaptable to the user needs.

In a world of openclaw, codex, claude code/cowork, manus, and other agentic systems, it’s becoming clear that the future of software has to be API-first, but also enable human interaction for verification, collaboration with agents and people, and working on the output. It’s generally been the case that software was built for people first and foremost, and then APIs are exposed for other systems to connect into that tool or data. But if we imagine a world where AI agents are doing 10X or 100X more work with software than people, then this paradigm is flipped. Software becomes API-first, with ways of having humans be able to work effectively with the agent, either through a UI as relevant, or chat. If you’re not API-first, then you’re nearly DOA to agents.

Silicon Valley gets all the Series A funding. But have you seen what’s being built in New York? Here are the companies I’m closely watching …make sure to keep your eye on them: @TennrOfficial – AI automation for healthcare referral workflows processing 10M+ documents/month @outtake_ai – Agentic AI that detects and removes phishing attacks and identity fraud @ModelML_ – AI that generates pitch decks and investment memos in under 3 minutes @Traba_Work – Labor marketplace for light industrial temp workers with 95%+ fill rates @FINNYAI – AI prospecting platform for financial advisors with automated outreach @flip_cx – Voice AI automating 90%+ of customer service calls @retirable – Retirement planning platform with fiduciary advisors for middle-income retirees @tidalwave_ai – AI mortgage platform with instant pre-approvals via Fannie Mae/Freddie Mac @hifortunahealth – Medicaid enrollment platform reducing coverage churn by 15% @warpdotdev – AI-powered terminal for modern software development workflows @draftwiselegal – Contract intelligence platform using firm precedent and AI @ExtendHQ – Extended warranty platform resolving 98% of claims in 90 seconds @rallyuxr – User research CRM automating recruitment and scheduling @crosbylegal– AI law firm reviewing contracts in under an hour @AmbrookAg – Farm accounting software with Schedule F integration Let me know if I’m missing any!

@BohanZhangOT @PostgreSQL @OpenAI @BohanZhangOT which ORMs are being used currently? Are schemas updates handled by any framework or an in house solution?

@PostgreSQL has long powered core @OpenAI products like ChatGPT and the API. Over the past year, our production load grew 10× and keeps rising. Today we run a single primary with nearly 50 read replicas in production, delivering low double-digit millisecond p99 client-side latency and five-nines availability. In our latest OpenAI Engineering blog, we unpack the optimizations we made to to scale @Azure PostgreSQL to millions of queries per second for more than 800M ChatGPT users. Check out the full post here: openai.com/index/scaling-…

Wth happened here today, my feed is full of ads… @X don’t need that

2025 was the year accurate intelligence became cheap and extremely powerful, giving engineers an Iron Man suit to do more than ever before. 2026 will be the year that AI leverage separates great engineering teams from the rest. So how do we prioritize AI leverage at @Traba_Work? Enter Roomba Rumble. Roomba Rumble is our way of celebrating engineers who maximize “Roombas” — background agent–led PRs that require minimal attention but have massive impact. Here’s some things we accomplished with Roombas in Q4: @TanzirisHere decommissioned our legacy event handling service, identifying migration targets and methodically working through them week-over-week. @sumeetbnsal replaced our legacy test infrastructure, cutting our CI times in half, and built Claude sub-agents to support our infrastructure. @Solidous became the first responder for operators' feature requests. Asks that once sat in a backlog now get shipped before others on the team even see them. @MaceoCk built a data analyst agent that generates research reports directly in Slack, turning guesswork into informed decisions. Other side quests have spanned the gamut: root-causing and patching noisy alerts, optimizing sluggish endpoints, even a Slackbot that randomly pairs teammates across functions to keep the culture tight as we scale. How did we celebrate them, you ask? With a turnover chain, of course. At Traba we celebrate the engineers who choose ambitious quests and see them through. Ultimately, we’re building a culture where force-multiplying AI leverage is owned by every engineer. If you want to apply frontier AI to the supply chains that stock your fridge, deliver your packages, and fill your prescriptions, come join Tanzir, Sumeet, Moreno, Maceo, and the rest of the team. Careers page below.

@SherryYanJiang @amix3k This

@amix3k might be unpopular opinion but 99% of cursor's users aren't using cursor yet so rapid iteration is far more important than just pleasing the early adopters im happy to take the friction so they can build the most kickass product

Cursor feels like a company without product management. Just devs and designers running seemingly random experiments with little higher-level strategy. I use Cursor a lot, and it is becoming a very frustrating experience: — The interface is cluttered with tiny icons that get shuffled around every week — Are they building an agent or an editor? Pick one. This hybrid approach, where they pivot weekly, is frustrating — “Everything and the kitchen sink” isn't a great way to manage a product. A better approach is focusing on the core, and make that great instead of changing/adding useless things — Focus on stability. People use this for real work. Running into bugs and issues daily isn't the experience anyone wants Compare this to Claude Code: one text box, clear identity as an agent. Simple and focused. It's no wonder it's growing a lot.

Gemini 3.0 just refactored my entire codebase in one call. 25 tool invocations. 3,000+ new lines. 12 brand new files. It modularized everything. Broke up monoliths. Cleaned up spaghetti. None of it worked. But boy was it beautiful.

@mike_sheb Just the beginning!!

While the rest of tech’s been busy trying to replace American workers, our AI platform has created 685,835 real work opportunities to real human beings. Our product learns from every interview and every shift, connecting workers to better opportunities at the essential businesses that keep our country running. Here’s how we’re doing it👇

@mike_sheb @Traba_Work 996 looks quite nice

@rotciVArt Joga Death Stranding

uma coisa que eu sempre percebo jogando jogos antigos é que se perdeu a noção de importância do vazio nem todo momento você precisa ta fazendo algo, precisa ter uma missão, o que faz um mundo parecer gigante e a ação impactar é o silêncio entre as partes do jogo

@dieegosf Graphite

Alguém conhece uma alternativa para review de Pull Requests que não seja a UI do Github web? Está cada vez mais insustentável a lentidão que é a UI dessa página. O React Compiler ou um vídeo da Rocketseat sobre memoização já salvaria essa página.

Inter didn't get the Flu vaccine.

Group B of the Club World Cup