Manuel S. Martone

We're moving from "automate the task" to "delegate the workflow." Agentic engineering isn't a buzzword, it's a new discipline. The engineers who get this early will define the next decade.

@fulhadev I know exactly what you mean because it happened to everyone out there (I even id they don’t admit that). The problem is context engineering is underrated as software desigb was underrated for years being expensive Now bad context can kill your pocket, that’s the point

@codebrainr ours grew to 800 lines. half is failure-mode catalog: additive bias, mock self-test, tool overlap. agents read it before every ticket. cuts re-explain time + drift. spec + post-mortem in one doc. context discipline > prompt cleverness.

A good CLAUDE.md file isn't optional, it's the difference between an AI that codes to your standards and one that confidently makes a mess. Context engineering is the new senior dev skill.

@RaminNasibov StarCraft

Anyone who used a computer between 1985-2010. What’s the one game you still think about?

Your side project is underrated career leverage. It shows range, initiative, and real-world execution. A GitHub repo ships louder than any bullet point on your résumé. Don't sleep on it.

The simplest advice possible almost nevertheless heard? Validate your idea with real people before writing a single line of code. A 10-minute conversation beats 10 weeks of building something nobody asked for.

Hot take: engineers resisting AI coding tools aren't protecting quality, they're protecting comfort. The ones quietly using AI are shipping 3x faster. Fear is not a strategy.

Promote your side project by leading with the problem, not the features. Nobody cares that you "used Next.js with edge functions." They care that you saved them 3 hours a week. Lead with pain.

That side project you abandoned in 2019 because it was "too complex"? Revisit it with an AI-first lens. The hard parts are now table stakes. Your old ideas deserve a second look

Your CLAUDE.md file is your team's constitution for AI coding. Skip it, and you're handing a junior dev zero context and hoping for the best. Write it once. Save yourself hundreds of corrections.

Autonomous Engineering isn't about replacing devs, it's about multiplying them. One engineer + well-orchestrated AI agents = a small team's output. The leverage is real. Are you using it?

Always validate before you build. Talk to 10 real users before writing a single line of code. Their words will save you months of building the wrong thing. Opinions are free. Rewrites aren't.

Hot take: "AI can't write good code" usually means "AI exposed gaps in my architecture." The engineers thriving right now aren't the best coders, they're the best context engineers.

Remember the mantra: Your MVP doesn't need a settings page, 10 OAuth providers, or push notifications. It needs ONE core feature that solves ONE real problem. Ship that. Everything else is procrastination dressed up as productivity.

@mitchellh @matteocollina (Í was writing you can't believe it) how much is this true? Every company, every team that I've dealt with has got quite the same psychosis, because AI needs to be everywhere, or there is the opposite psychosis that AI and agent coding cannot be trusted.

I strongly believe there are entire companies right now under heavy AI psychosis and its impossible to have rational conversations about it with them. I can't name any specific people because they include personal friends I deeply respect, but I worry about how this plays out. I lived through the great MTBF vs MTTR (mean-time-between-failure vs. mean-time-to-recovery) reckoning of infrastructure during the transition to cloud and cloud automation. All those arguments are rearing their ugly heads again but now its... the whole software development industry (maybe the whole world, really). It's frightening, because the psychosis folks operate under an almost absolute "MTTR is all you need" mentality: "its fine to ship bugs because the agents will fix them so quickly and at a scale humans can't do!" We learned in infrastructure that MTTR is great but you can't yeet resilient systems entirely. The main issue is I don't even know how to bring this up to people I know personally, because bringing this topic up leads to immediately dismissals like "no no, it has full test coverage" or "bug reports are going down" or something, which just don't paint the whole picture. We already learned this lesson once in infrastructure: you can automate yourself into a very resilient catastrophe machine. Systems can appear healthy by local metrics while globally becoming incomprehensible. Bug reports can go down while latent risk explodes. Test coverage can rise while semantic understanding falls. Changes happens so fast that nobody notices the underlying architecture decaying. I worry.

Validate your idea with real humans before writing a single line of code. Seriously. One conversation > 200 hours of building something nobody asked for. Talk first, build second. Always.

@bourneshao Exactly! I’d say the important things are effectiveness and structure. Do you know treating it as a router for other meaningful files is way better than having one with details and rules?

Your Claude.md file is your agent's brain. Write it badly and your AI coding assistant will confidently do the wrong thing, every single time. First step to reliable AI-assisted dev: write a great CLAUDE.md. Everything else follows.

AI Agent Automation is the new gold rush, and most devs are still panning by hand. The engineers building agentic workflows today are the ones writing the rules tomorrow. Pick a side.

Hot take: most "AI won't replace me" engineers are protecting their comfort zone, not their craft. Resistance to AI adoption isn't professionalism. It's fear dressed up as principle. Adapt or get comfortable being disrupted.