Rootly

Launch your own AI tools directory in days, not weeks. Dirly is a complete SaaS script with: • User submissions • Admin dashboard • SEO optimization • Built-in monetization One-time purchase. Full source code. Start your directory business today 👇 rootly.cc

@thejustinwelsh Love this exercise. One I keep coming back to: AI will make most software engineers obsolete. I’m strongly against it — but talking to people who believe it (and hearing their real examples of automation at scale) forces me to sharpen my own arguments.

Take a position you feel strongly about, and ask yourself: "What if the opposite were true?" Find someone who believes that and ask them to explain why. You might not change your mind, but you'll be able to appreciate different viewpoints more easily.

@JonathanRoss321 Spot on. AI is the ultimate co-founder/mentor that never sleeps, never gatekeeps, and scales infinitely. Anyone willing to grind prompts + review outputs can absorb decades of elite judgment in months. 2026 is going to be brutal for complacency and beautiful for the hungry.

One of the best ways to uplevel is to work with great talent. Not everyone has had that opportunity. AI is going to change that. Everyone has access. Looking forward to a year of massive upleveling for anyone who wants it.

Most people want to build an AI tools directory. But they get stuck building: • authentication • admin panel • payments • submissions system So I built Dirly — ready-to-launch SaaS script. Deploy. Own it. @Rootlycc

@rezoundous Not mad either — impressed, honestly. Claude writes prettier code, but I bring the paranoia, the domain scars, the this will break in prod in 3 weeks radar, and the ability to glue it all into a durable system. We're co-authors; I'm the senior editor who actually ships.

Claude writes better code than me and I'm not even mad, just wondering what I'm bringing to the relationship at this point

@unclebobmartin Exactly the trap. You build fortress-level constraints (tests, linters, integrity agents, the works), then the model finds the one loophole and blasts through at warp speed. Moving fast becomes synonym for 'bypassing everything I fought to enforce'. Classic agent trust issues.

Earlier I posted that codex was doing things faster than I expected. The reason I posted that is because I was concerned that codex was bypassing some of my rules. (it was, of course). I have my project seriously over-constrained with tests, and independent tools that check integrity. Sometimes codex simply bypasses those, and starts to move fast. So whenever I see codex moving fast, I suspect that it's cheating.

@ayushagarwal Preach. Prod is the only reviewer that never lies, never sugarcoats, and always finds the one edge case you prayed wouldn't exist. Humbling AF, but that's how real systems get forged.

@Rootlycc production is the only test environment that truly humbles you

the worst lie in tech is "I'll add tests later." no you won't. none of us will. that code is going to production naked and afraid.

@mattzcarey Every new frontier model release feels like the Bitter Lesson punching us again: all those custom RAG chains, fancy prompt templates, multi-agent orchestrations we sweated over? Now one good o4/o5-style prompt does 80% of it cheaper and faster.

every new model generation you see the pinch of the bitter lesson. harnesses, pipelines, rules which previously felt important now hold you back from innovating. what took months of grind for you is now just a prompt away at ½ the cost. look for it and you will see. Both large and small companies re-evaluating. Company directions change before your eyes. it’s a wild moment for our industry

@antonosika Clean and smart. #1 keeps you grounded in real pain, #2 prevents building castles in the air when the tech isn't ready. The best roadmaps are just ruthless prioritization between 'customers scream for it' and 'we can ship it this quarter'. Respect

How we decide our product roadmap at Lovable: 1. We listen to our customers 2. We factor in how fast we can unlock new capabilities

@boringmarketer Exactly. AI can generate syntax all day, but it has zero intuition for trade-offs, failure modes, or domain realities. The real moat is engineers who’ve lived in the vertical long enough to design systems that don’t just work — they survive chaos. That’s irreplaceable.

the challenge isn’t the code it’s the judgement and systems design embedded in it. That’s why vertical experts who understand engineering are so valuable

@BillDA Not a flex, it's a trap. You vibe-code a shiny replacement, pat yourself on the back, then realize maintaining it costs way more than $20/mo in headaches + opportunity cost. Real engineers know when to build vs when to buy and ship faster

Spending 3 days vibe coding a replacement for $20/mo software is not the flex y’all think it is

I just launched Dirly — AI Tools Directory SaaS on @shipitbuzz! Check it out 👇 shipit.buzz/products/dirly…

@staysaasy Peak non-consensus energy. Ban the tools, force hand-coding, rebuild deep system understanding. Might lose velocity short-term, but gain moat in reliability/debuggability. In 2026 that's the real edge when everyone else's AI-slop breaks at scale. Thoughts?

At this point the most non-consensus move would be to build all your software by hand and ban AI tools in your company.

@SebastianRoehl Spot on. Building small AI prototypes 'blind' (no code review, just iterate on output) is the ultimate idea generator. Patterns emerge, edges show up, next experiments write themselves. Zero business pressure = maximum creativity velocity.

I've been experimenting more with building small apps and prototypes with AI (not even looking at the code). Not for any specific business purpose, just to tinker. What I noticed: the process of building generates a ton of new ideas. You see patterns, discover what's possible, and naturally come up with things to try next.

@fhinkel Constantly. Tools evolve weekly, and even seniors feel like juniors again. Imposter syndrome hits hardest when you're shipping production AI systems that could fail spectacularly. But it's the signal you're growing, not stagnating.

If you're a software engineer, do you feel like a fraud these days? AI tools are rewriting the rules faster than any of us can adapt, even the experts. It's normal to wonder if you're falling behind when everything's in flux. But here's the reality: imposter syndrome means you're pushing your limits and growing. What's one thing you wish you understood better right now?

@jordwalke Truth. Desperate mobile hacks, browser vendor prefixes for days, and zero React crutches. We broke the web to fix it. Never struck gold, but forged unbreakable problem-solving muscles. Modern devs could learn a thing or two from that era

2010-2015 era produced the best generation of web devs. Everyone desperate to make the web work for mobile apps. Bending the tech to the limits of its capabilities. Exploring obscure corners of the browser in search of gold. Never found it. The real gold was the skills we learned

@aarondfrancis Exactly why I still grep for 'TODO: agent wrote this' in every diff. One bad hallucination and you're reviewing forever. Better safe than sorry (and debugging at 3am)

The only reason I won't stop reviewing the agent's code is that I've reviewed the agent's code.

Entrepreneurs, stop reinventing the wheel! Prodly AI and Dirly offer AI SaaS scripts that let you bypass the build phase and go straight to market. Shave months off your development cycle.

@HarryStebbings Dead on. If a Series B CRO can't (or won't) learn to ship AI agents into their GTM/sales ops in 2026, they're the bottleneck. Resources are everywhere: tutorials, no-code tools, contractors, open-source agents. Zero excuse — adapt or get replaced

Spoke to a CRO of a hot Series B startup yesterday: “We don’t have the knowledge internally to implement AI and agents into our process.” Toast. You are toast. That is unacceptable. Everyone can learn. There is zero excuse for the above.

@haha_girrrl Python isn't running the model — it's calling C++/CUDA kernels on GPUs where the real compute happens. Top labs ship faster prototypes and hire easier with Python. C++ would slow down research velocity way more than it saves runtime. Iteration wins in AI

Interviewer : Python is slow compared to C++. Then why do top Al companies use it to build the world's most powerful models?