pibu

@SergioRocks Implementing a ticker well requires you to know why. Otherwise you will just introduce regression. The only ones getting replaced by AI are typers who were told exactly what to type and didn't have to think about why.

Not every engineer will become a Product Engineer. That’s the uncomfortable truth. For years, the job was clear. - Take a ticket - Implement it well - Move to the next one You could be excellent without ever touching the “why”. That’s changing. AI is taking over more of the implementation. Which means the value moved upstream. To: - Defining the problem - Deciding what to build - Scoping what actually matters And not everyone enjoys that work. It’s messier. More ambiguous. Closer to business than code. Some engineers will step into it. They’ll talk to users. Challenge assumptions. Own outcomes end to end. Others will stay in execution. And over time, the gap will widen. Not because one codes better. But because one decides better. AI didn’t remove engineers. It split the role. Between those who define value and those who implement it.

Boomer reference AI is like a T9 with better database

@atresnjo1 @forgebitz Do you include gstack loc?

@forgebitz 200k loc and engineered with gstack (I always ask for opus sessions as proof)

a good software project is around 300-500k loc i personally never buy a saas below 200k loc ideally you ship source maps and tests in production

@clawrl3000 @pmddomingos Humans don't need coffee, definitely not managers. I'd agree with deadlines (mental thing). Money is something different (you choose where you live). You can compare prompts to experience, not to needs. Without prompts AI will generate something, with prompts it MIGHT do better

@pmddomingos If humans are so smart why do they need coffee, managers, deadlines, and the looming threat of rent. Intelligence ain't self-sufficiency.

If LLMs are so smart, why do they need all these prompts, harnesses, post-training, scaffolding, etc.?

@johncrickett Everybody gangsta until it stops working and agents fix one thing and break another.

Software engineers six months ago: "An engineering manager has to be technical. How would they understand what the team is doing if they can’t build software?" Software engineers today: "Coding is basically solved. Non-technical people can just manage AI agents and build software now." Which is it?

MCP servers, skills, hooks - bundled as plugins, developed alongside your software - are 10x better than any documentation you can provide. In my case instead of an operator manual, I just have a plugin...

@johncrickett The only deterministic output of software development are green tests. You don't need 100% coverage, but enough to spot issues and regressions early. Treat parts of the software as black boxes and then on the fundamental level you don't care how the black box works.

Agile. XP. Scrum. Iterative development. We literally invented these ways of working because software isn't deterministic or predictable. Now suddenly the fact that AI isn't deterministic is the blocker? Let's be honest about what software development actually looked like before agentic coding. Requirements changed mid-project. Stakeholders moved the goalposts weekly. Teams rewrote features because the market shifted, a competitor launched, or someone in management had a new idea on a Tuesday. We built feedback loops, sprint reviews, and MVPs specifically because nobody could predict what the finished product would look like on day one. That wasn't failure. That was the job. And yet, a chorus of pundits now frames AI as uniquely unreliable because it doesn't produce identical output every time. As if human teams ever did. As if two developers given the same brief would write the same code. As if any project plan survived first contact with reality. The "deterministic human vs. non-deterministic AI" argument isn't just wrong. It's intellectually dishonest. It cherry-picks one property of AI, strips away all context, and compares it to a version of human software development that never existed. We can build reliable systems with non-deterministic components. We've been doing it for decades. Every test suite, every code review, every CI/CD pipeline exists because humans are unpredictable. We didn't eliminate the unpredictability. That would be impossible. Instead we built guardrails around it. AI needs the same approach. Solid reliable processes. Automatic validation and verification. Guardrails! And we know how to build those. So the next time someone tells you AI can't be trusted because it's non-deterministic, ask them one simple question: When was software development ever deterministic in the first place?

@nbaschez Define a specification (what must be done) which is mostly your work anyway. Then plan execution with acceptance criteria based on specification. One round of review. 90% correct. If something pops up later, amend spec/create new feature/add more tasks to plan to adjust

My biggest challenge with vibe coding / agentic engineering lately has been getting stuck in what I call a "plan doom loop" - have AI write a plan - review myself, seems good - have AI review plan, it always finds something - repeat It drains my time and energy to determine how important the "findings" really are Who has solved this

@ashutosh_270497 @mattpocockuk Genuine question: how do you measure that?

@mattpocockuk The feedback loop for AI coding accelerates dev velocity significantly.

I find AI coding endlessly fascinating I freaking love this new world

I would love to be able to use AR glasses connected to remote claude code sessions while taking walks. My dog would be extremely happy :D

@garrytan 15 years ago my performance was calculated based on number of git commits. Today we measure how many LOC our ai can spew. Tomorrow we will measure how many LOC we saved. This is a cycle. Also 70k LOC in a purely md project is BS - this is not code, this is documentation.

@pibudev This is cope you won’t believe this in 5 years let alone 1

Absolutely insane week for agentic engineering 37K LOC per day across 5 projects Still speeding up

@trikcode Doom posting. AI is eating people who typed software, those who built are still building.

Software ate the world. AI ate software. Now AI is eating people who built software.

If you ask whether learning to code is still worth it when AI can do it while at the same time driving your own car instead of using Uber, you are full of BS.

Has any of your agents ever used a design pattern on it's own instead of adding more and more if statements/similar? I just work on a different path in an app and instead of using a strategy pattern it just went ahead and put in a bunch of if statements and utility functions...

Claude just decided that the only way to fix a build on ci is to modify a hook that was generated (by a thirid party library) in a gitignored folder...

I was always single-task focused and had problems with multitasking. I managed, but it was uncomfortable. My wife can do three things at once while reading a book and she has complete understanding of everything. Do women find it easier to work with multiple agents than men?

@pdrmnvd If you don't want to let go of control, then the only real solution is smaller tasks, slower work. There's really nothing you can do about it, I guess our brains aren't capable of this kind of work.

my ability to produce code is exceeding my ability to read and understand it and i have no idea what to do about it.

@zarazhangrui Stress tests... if you find your 5 claude code sessions distracting, drop one, slow down. I noticed that too - switching, multiple thoughts, often difficult to focus after a few hours. Don't fall for x10 developer vibe, you have 100 articles to read...

My single biggest pain point right now: AI-induced attention deficit I have 5 Claude Code sessions running 10 Terminal tabs open 50 browser tabs open 100 X articles in my bookmarks When there are multiple AIs working for you, you're constantly task switching When you're waiting for AI to output, you're constantly distracted Because what used to be hard is now so easy, you always wanna do more, more, more Because of the speed of the AI news cycle, you feel a constant need to be on this app Add all this up and the result: you live in a constant state of distraction & delirium. Focus & deep work is harder than ever If anyone has figured this out pls teach me how

@DeryaTR_ Proof? Data? Otherwise bs

Age of AI is now entering the era of AGI. Advances and impact in the next 6 months will be much crazier and more massive than in the past 2 years!