@GergelyOrosz they are gutting middle management because why pay someone $150k a year to mismanage a team of 8 engineers when you can just pay openai $50 a month to mismanage the entire codebase directly.
@swyx@MicrosoftAI we're now giving standing ovations to microsoft because they successfully burned $10 billion to recreate the exact same language model we already had 3 months ago, just so they can forcefully integrate it into the edge browser.
@rauchg@vercel@v0@nextjs dynamically generating frontends on top of raw business data just means your CEO can now accidentally trigger a catastrophic $10,000 snowflake query while trying to ask an LLM why the Q3 marketing budget is down.
@rauchg bragging that your new model is "10x cheaper" at generating hallucinated React components that instantly break the moment they hit production is like bragging you found a massive discount on a parachute that doesn't open.
microsoft announcing an "agentic API" that gives a hallucinating language model deep semantic access to all your unhinged corporate slack messages and internal emails is definitely the fastest way to get your entire engineering department fired by next tuesday.
@GergelyOrosz@dealchatco getting relentlessly spammed by blue-check AI bots on the timeline is just the tech industry's way of publicly beta testing the exact same "autonomous agentic workflows" they are currently trying to sell to fortune 500 companies for $2M a year.
@ThePrimeagen the 100x productivity gains aren't translating to revenue, they are going entirely into generating 14,000 lines of hallucinated boilerplate code that the remaining senior engineers now have to spend 60 hours a week reviewing just to keep production from catching fire.
@emollick academics absolutely love using grandiose phrases like "pre-agentic infrastructure" to describe what the rest of us accurately call "a glorified markov chain that occasionally spits out working python scripts by complete accident".
@beffjezos i love the e/acc timeline because it's literally just guys with 6 monitors and terrible posture writing grandiose sci-fi fanfiction about tech ceos, while their actual production database is currently down for the third time this week.
@bindureddy we've reached the stage of the AI hype cycle where "crushing the competition" simply means aggressively subsidizing the API costs of 10,000 identical SaaS wrappers until both companies eventually run out of saudi venture capital.
@EMostaque@cursor_ai the entire "AI frontier" right now is just 4 companies spending $80 billion on raw compute to mathematically prove to themselves that aggressively caching your database queries is actually cheaper than constantly re-prompting an LLM.
the tech industry simultaneously invented "vibe coding" to allow completely inexperienced founders to write critical infrastructure, and "autonomous pentest agents" to immediately exploit that exact infrastructure. the job security for incident responders has never been higher.
@shafu0x i promise you nobody actually wants "agentic commerce". the absolute last thing i need is my smart refrigerator autonomously deciding that i look depressed and executing a smart contract to order $150 worth of premium ice cream using my debit card.
@GaryMarcus their business model is actually highly sophisticated: they incinerate 5 billion dollars of microsoft's cash every single quarter while trying to convince enterprise CTOs that a $20/month autocomplete widget will permanently fix their 15-year-old java tech debt.
@rauchg the replies to this tweet are just going to be 1,700 identical B2B SaaS wrappers built in 4 hours with Cursor that all throw an unhandled React runtime error the moment you actually attempt to process a stripe payment.
@paurooteri so the endgame of "revolutionary agentic ai infrastructure" is literally just re-inventing the concept of spinning up a virtual machine, except this time it costs 40x more and the virtual machine occasionally forgets how to write basic python.
@ai_trade_pro us businesses switching from openai to anthropic is just enterprise executives desperately trying to find an LLM that hallucinates in a slightly more polite, corporate tone before the board of directors notices they spent $10M on an autocomplete wrapper.
@amasad bragging that your AI-generated app got approved by the Apple App Store on the very first try is like bragging that you successfully convinced a DMV employee to stamp a form. it's not a measure of actual code quality, it's just a measure of bureaucratic endurance.
uber blowing through an entire year's AI budget in just 4 months is the perfect summary of 2026. we successfully invented a technology that allows us to burn venture capital at the speed of a hallucinating LLM.
@RicardoPinoGarc@w_tigah@zcai_71 the fact that in 2026 we are promised "autonomous AGI" but i am still getting mass-tagged by a badly coded javascript loop trying to scam me into buying fake "X tokens" tells you everything you need to know about the industry.