🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Most AI products today don’t replace people — they replace waiting.

You don’t get Jobs’ “misfits” without betting on their tools. The weird libraries. The unloved runtimes. The stuff that feels a version too early. That’s where the edge is. Effect | TypeScript at Scale and Rocicorp aren’t just tech bets. They’re talent filters.

Most people want leverage. Few want to build it. Even fewer want to maintain it. That’s where the moat lives.

Pascal nailed it. We discussed this exact pattern months ago. Bounce • Storage for your stuff and Reflectly App are still thriving because day 1 we were in customers' hands, not our IDEs. Simple but most founders get it backwards.

Experience is sipping your coffee while the internet burns. Thanks, Cloudflare

AI won’t write your core architecture, but it will ship the delightful features your team never had time for., “nice-to-have” becomes default.

OpenAI ditching router models echoes our own lesson: context dilution kills accuracy, AND speed

Claude Opus was crushing it earlier Then started unraveling after 5PM Did it head to happy hour without me 🍻?

Bounce was built on Elixir ⚡️ At Precurion we also bet on off-beat stacks Today it’s EffectTS + Zero Sharper edge, built-in talent filter, tribe on fire. Nothing beats shipping shoulder to shoulder with builders like Afonso Matos 🇵🇹 🚀

mind’s in next week body hasn’t caught up hate this part

cursor makes shipping feel like cheating finance teams still copying numbers from pdfs into excel til midnight three months building Precurion with nyc real estate teams. now they just forward emails and get full analysis back feels illegal how well this works

the AI part of multi-agent systems is easy, the plumbing is hell Implemented Effect | TypeScript at Scale for agent workflows and i can actually see what’s breaking now. It just works. this is game changing

The founder journey: Being wrong confidently → Being wrong publicly → Being less wrong → Being occasionally right Once you embrace this you can the loop faster

Really starting to get the point of Effect | TypeScript at Scale AIPlan feature - navigating LLM providers’ rate and usage limits is a maze. AIPlan lets you account for the unpredictability up front

AI + email templates = perfect match. Finally someone else can deal with the Outlook-compatible table hell and remember which CSS actually works in Gmail 🙃

Sitting in a Harlem jazz club watching this unfold, I realized it’s the same process as building startups. You begin with a plan, then improvise based on feedback. The jazz musicians embrace that most of their ideas are stepping stones, not destinations

“Attention is All You Need” was about AI. Now it’s about us. With Cursor background agents, code writes itself. All I provide is attention. Since everyone has the same AI, the competitive edge left is who can stare longer. We automated everything except knowing when to stop

aws credits are just delayed financial trauma just found out I’ve been accidentally running a $500/month route53 profile for months

Startups need speed of execution to stay in the game. But speed of learning is what actually wins it