Krishal

claude adding /goal is not a small CLI update. the race is moving from: write this code to: keep working until this condition is actually met that’s a different kind of coding agent less chat more task runner codex vs claude is getting interesting again

@shadcn this is true for most product apps. users don’t care if your dashboard looks unique. they care if they can find the button, finish the task, and not wait 4 seconds for a fancy animation.

Unpopular opinion: I don’t care if most web apps look the same. All I care about is whether it does what it says and does it fast. Make it fast. Make the UX obvious. Put the right things in the right place and little to no animations.

the new chatgpt feels different now not just smarter answers more like it understands the way i ask, the kind of work i do, and the context i usually forget to explain model + personality + personalization is a much bigger shift than people think

@sama optimize for merge ready PRs not just working code. better repo-wide context, stronger test generation/debugging, project-style consistency, clearer commit/PR summaries & explicit risk/edge-case notes before opening a PR

what would you most like to see improve in our next model?

We need value & knowledge... That's it...

@justbyte_ Disagree 👎

claude is still better than codex at coding, if you just excuse the token limits.

Vibe coder vs engineer is not “uses AI vs doesn’t use AI.” Both use Codex. Both ship faster. The difference: Vibe coder asks: “does it work?” Engineer asks: “why does it work, where can it fail, and can I maintain this later?” AI didn’t kill engineering. It exposed it.

@haider1 Huge agreed 💪

gpt-5.5 is pretty solid so far gives sharp answers, pushes for accuracy, refines ideas well, and is useful for brainstorming it also handles language nuance better, makes fewer mistakes, and works well for agentic coding in codex > set reasoning to high in codex > set reasoning to extended in chatgpt

@Aditya_181105 Exactly dear..

@srekrishal By giving clear, small, and specific instructions so the AI only changes what’s needed

AI coding agents are useful, but sometimes they turn a small bug fix into a huge code change. The real skill is giving them a tight task so the review doesn’t become harder than the fix. How are you keeping AI-generated changes small?

@ayesha_fatiima This is so true. The earlier you understand fundamentals + consistency matter more than chasing every new tool, the less confused your dev journey becomes. Most of coding is just patience, debugging, and not quitting when things break.

Things every developer should realize early: • You do not need to learn every framework • Fundamentals outlive trends • DSA improves thinking, not just interviews • Projects matter more than certificates • Consistency compounds quietly • Bugs are normal, not failure • Documentation is better than most tutorials • Tech changes fast, learning ability matters more • Building teaches faster than consuming content • Comparing your journey kills progress • Communication is also a developer skill Coding is not about memorizing syntax. It’s about figuring things out without giving up.

@Aditya_181105 Cloudflare for clean pricing and no upsell noise. Namecheap is also solid if you want a simpler beginner-friendly UI. GoDaddy feels like fighting popups before buying the domain 😅

Hey devs, which is the best place to buy a domain? >GoDaddy >Hostinger >Cloudflare >Namecheap

What’s the hardest part of backend development? A) Designing APIs B) Handling auth C) Scaling services D) Fixing production bugs

@LauraLunaTech Exactly. More like a portable trust layer. One verified identity, then other platforms can say: “yes, this is the same creator.” Way better than handing biometric data everywhere.

@srekrishal It's probably more likely that you get an account verified on a trusted platform and then prove it's you through the account at other places.

In the future, we will not only need labels like… Made in Germany, Made in India, Made in Korea, but also some kind of label or verification like… Made by a human.

@LauraLunaTech I hope not biometrics 😅 Probably a mix of platform verification, signed provenance, and creator history. Less “prove your body is real.” More “prove this work came from this trusted human account.”

Agreed, we are already seeing more people willing to pay for an individual's blog post, something that seemed unthinkable a few years ago (Substack), and also artist communities that gatekeep submissions! How do you think creators will be verified? Blockchain? Some kind of biometric data?

AI coding agents are getting good enough that the real risk isn’t bad code anymore. It’s giving them too much access. I’m starting to treat every agent like a powerful junior dev: branch only, scoped tokens, no prod keys, review before merge. How are you setting guardrails?

@frog_omo Exactly. Fast code review is the new superpower.

@srekrishal The new developer skill might become: “spotting subtle AI mistakes quickly.”

AI coding agents are getting good enough to build features. But I think most builders are about to hit a different problem: review overload. The bottleneck is no longer “can AI write code?" It’s “can I trust this diff without reading 900 lines?”

@Aditya_181105 This is the real backend roadmap. Frameworks change, but these concepts keep showing up once your app has traffic, failures, queues, retries, and multiple services talking to each other. Especially event-driven systems + tracing. That’s where things start getting real.

If you want to understand how modern backend systems are built, study these: >REST vs gRPC >Monolithic vs Microservices >Load Balancing Algorithms >Database Replication Models >Event-Driven Systems >Message Brokers (Kafka/RabbitMQ) >API Gateway Architecture >Service Discovery Mechanisms >Circuit Breaker Pattern >Distributed Tracing Systems

AI coding tools make it very easy to feel productive. More prompts. More generated code. More “progress”. But I’m not sure that always means better software. The real edge might be knowing when to stop asking AI and start shipping. How are you measuring AI productivity in your workflow?