Most backend devs are asking: "Which AI tool should I learn?"
Better question: "Can I turn a business problem into a system design without hand-holding?"
That's the skill AI can't replace. Most devs never built it.
#BackendDev#SoftwareEngineering
@claudeai mythos released today. the model everyone was worried about. can find what humans miss.
costs 2x opus. moves fast. doesn't replace security researchers but... it's not trying to.
if you're building infrastructure or products that matter, time to get familiar with this.
Anthropic says Claude is already accelerating its own AI development — recursive self-improvement is no longer hypothetical. We are living in the most important moment in tech history.
@ycombinator@steipete years of backend work and the projects that blow up on GitHub always have one thing in common — the core loop is so simple that a dev goes from clone to "oh wow" in under 10 min. openclaw nailed that. the idea matters less than friction to first experience.
@outofofficedaku This is absolutly devastating. Flying all that way just to be turned back is a financial and emotional nightmare. Unfortunately, EU immigrtion has become incredibly strict about verifying accommodation and ensuring the issuing country matches the primary stay.
@pharmeasyapp Spoke to your customer service at 08071324888. I was told delivery will take another 24–48 hours, despite the SLA promised on your website. What if a customer suffers due to this delay and broken commitment? Please resolve this urgently.
Complaint #679048
⚠️Trigger warning⚠️
Executives, VPs and Directors of the last decade are cooked…
The disconnect in 2026 between the executives between how work has happened historically and how work now happens with AI is striking.
The truth is you ACTUALLY have to work now. Everyone is a IC. From the CEO to the intern it doesnt matter. The organization of the future is IC to exec no middle management. There is simply no need for it. AI is advancing so quickly if the VP isn’t hands on they are deprecated in ~30 days.
The silver lining is there is unlimited opportunity for Gen-Z and millenials willing to step up.
@nutanc Every backend dev has that one scar from spending 3 months building something a $50/mo SaaS already did perfectly. AI just made that scar easier to earn faster.
@addyosmani From a backend lens: every agent is essentially a long-lived process — and most systems were never designed for that. State management, retry logic, timeout handling all need rethinking. More agents ≠ more output. Just more surface area for failure.
@addyosmani@googlecloud The backend challenges here are the ones nobody talks about enough: idempotency, checkpoint/resume on failure, and cost observability across multi-day runs. Long-running agents make infra costs wildly unpredictable without tight guardrails at the architecture level.
@dhh The "stable packages vs high-churn AI tooling" framing is exactly right. As a backend dev, environment drift across teammates is one of the biggest silent productivity killers. A single config managing Node, Go, and Python versions is genuinely underrated.
@GergelyOrosz From a backend lens, this pullback makes sense. Many teams bolted AI onto existing systems without redesigning for it. Now they're paying for inference, storage, and latency they never budgeted. The orgs cutting spend are finally asking: "what problem does this actually solve?"
@GergelyOrosz This resonates deeply. In 14 years of backend engineering I've seen the fastest teams burn out chasing velocity, then spend months untangling the shortcuts. The teams that shipped the most long-term value were the ones who paused to think before they typed.