Steven Liss

We’ve reached the point where my Claude is more deeply ad-pilled than me

@shinboson tbf, middle schoolers are high agency psychopaths with no concept of their own limits

@dunkhippo33 Most generational companies were founded within 12 months of being technologically feasible. The exceptions were network effects. But even Facebook had the wave of student laptop adoption to help vs MySpace before it.

An important question to ask before starting a business is "Why now?" As a founder, I used to think that was such a silly question, but now, having met so many super sharp entrepreneurs, you've got to believe that for every great idea, there's somebody who would have gone after it if it had already been an available opportunity five years ago. This isn't always the case. Sometimes there are great opportunities that have been overlooked for years, but often it is.

@tengyanAI There’s no compute shortage. The token limit is for your safety.

something i've noticed: AI agents create a weird new kind of burnout. esp for young people. a lot of ambitious 22 year olds are going to think the answer is simple: - spin up more agents - ship more code - sleep less - outwork everyone and for a while, it will feel incredible. you can keep multiple agents running, feed them tasks, review outputs, fix mistakes, make decisions, and keep the whole loop moving. the problem is that the work no longer drains you through typing. it drains you through judgment. More attention. More context switching. More verification. More decisions per hour. so instead of 8-10 normal productive hours, you might get 4-5 extremely intense hours before your brain is fully cooked. and you feel numb until you sleep properly and reset some of my friends are already burnt out. they don't say it out loud but i can tell. the agent can keep working 24/7. the human still has a hard limit

You could join the 10th legion tomorrow. 25 years, then a plot of land in Hispania with olives already bearing. Marry a centurion’s daughter who can run a household. Pour libations to your ancestors. Teach your sons to ride. Be buried by grandsons in soil you broke yourself, in a province at peace because men like you held the Rhine. But you won’t. You’ll keep your clerkship in Subura. You’ll tell yourself 30 is still young for the cursus honorum. And in ten years a Gaul your age will own the farm that was meant for you.

@svpino Counting “agents” is like counting CPU threads. I have 1-4 Claude Code instances running 0-40 subagents at a time. It’s not a useful metric.

I don’t believe people who say they are running “12 parallel coding agents”. Either they are lying for clicks, or I’m a complete retard who can barely keep up with a single Claude instance.

@krishnanrohit @emollick @hypersoren We shouldn’t accept the permanent underclass as inevitable. That’s permanent-underclass talk!

@hypersoren This entire discussion of permanent underclass is incredibly toxic

Worth observing what the labs/scaled players are actually doing -quietly setting Claude effort to medium -Cursor efficiencymaxxing The correct strategy is to counterposition against token consumption not be Cursor with -23% margins. Dylan just trying to sell his course

@TheIshanGoswami @mil000 He just wants to keep all the alpha to himself

@mil000 yoo king, would love to hear your feedback on what all you felt we are bad at and happy to drop credits for you to try exa once again

Exa genuinely sucks, please don’t use it.

@trishlaostwal @AnthropicAI @Uber @Instacart @Spotify @resy Claude will need some way of deciding which interchangeable services to surface to users… and those services may provide incentives like deals for users or even paid MCP integrations… but they will certainly not be called “ads”!

Inbox: @AnthropicAI, while staying away from ads, is expanding Claude’s “connectors” into everyday consumer apps-- @Uber, @Instacart, @Spotify, @resy-- positioning the AI to execute tasks like "finding a reservation, adding to a grocery cart, or identifying a flight."

@Bfaviero @signulll Airbnb in developing countries still has gems. - written from an Airbnb in southern France

@signulll AirBNB. I know so many people now that prefer hotels.

@ericfranchi It’s like Taboola vs Outbrain competing for inventory with unprofitable CPM guarantees… except with more venture money chasing fewer publishers.

There are a bunch of AI chatbot-focused ad networks chasing the same AI chatbot app inventory.

@shinboson average adtech professionals weighing in at negative 300 IQ 🧠

@hypersoren Welcome back to NYC, the human capital!

I am moving back to New York. I spent the last few years in Miami Beach. I moved for family but I broadly felt—if all predictions of AGI came true; if we had agents that could do arbitrary tasks—beach property would be valuable. I’ve been disappointed with how early stage AI has developed. Despite grand narratives > THIS IS AGI investment in kind hasn’t followed. The SF Taste Class dooms aimlessly about what we’ll all do when Claude does most intelligent work. It sounds like a joke but Miami already lives in this future. The city is not pretentious about intelligence—perhaps entirely the opposite—but it’s undeniable that Miami is one of the highest growth and dynamic cities in America. The city is deeply American and lives its true self 24/7—party boats, cars, beautiful people, money, rich immigrant culture. Miami dominates @alexolegimas Relational Sector—human intensive, provenance rich service where human-aspect of services lead. New York probably too. SF—not so much. Plainly I don’t think people are thinking clearly about economics in this new era; acting as if AGI invalidates underlying economic principles. I’ve heard rumors some at Anthropic want Anthropic branded everything, all run and organized by a Constitutional Claude Commune. I don’t want to live in this future. Beyond SF’s incredible natural beauty, its cultural and political aesthetics are dismal. The AI narrative is now just Permanent Underclass For All but labs Who Will Charge 90% Margins. I reject this. I reject these aesthetics. I reject cosplay. If we really are ushering in a magic machine god made of sand we must take everything seriously across all domains of life. When the cycle started we were Accelerating. We were on our way to Abundance. Now we twist narratives to accumulate LP dollars to rent seek compute replacement for human labor. How profoundly depressing and unimaginative. Also—deservedly—not durable. So I’m headed to New York to build Technologies Of Abundance. See you downtown. JPMorgan Finishers better up the pace on the Central Park hill.

@scottastevenson SC pro-tip: assign your terminal windows to control groups to toggle between agents quickly I use ⌘1,2,3,4,5. Add more as your macro improves.

APM seems like a very unintuitive and pointless metric to most people. But here’s the magic: when you’re focused on increasing APM, it trains your brain to think faster. It creates a constant monitor in your brain pushing you to make decisions and move.

Starcraft is an incredible foundation for entrepreneurship. It wired my brain at an early age to think about ROI, APM, macro, micro, the benefit of keeping your minerals at $0. It is no coincidence that many of the great founders of this generation have played this game. Here is my APM (actions per minute) tracker that I use for playing the startup game (had it closed for a couple days -- I still APM on the weekend). One of the things you learn playing SC is that there is a base level of APM needed to be successful. The speed at which you can plan and act on a computer is a lot faster than most people realize. It's not everything. But if you aren't hitting a minimum level you will just get overwhelmed and get crushed.

@garrytan AI coding is as fundamental to knowledge work as literacy. Not a weird take.

I came back to code because AI made it possible for me to build at a level I couldn't before. I'm not coding despite being CEO of YC. I'm coding because this is the most important technological shift since the internet and I'd be an idiot to experience it from the bleachers. I'm 45, running the most important startup institution in the world, and I can ship production software at 2am. That's not a distraction from the job. That is the job understood correctly.

@signulll you all still using roads?

in the ai era, if your “roadmap” is > 30 days, you’re simply ngmi.

@tanayj @eastdakota Open internet = agentic internet. These agents aren't living in the walled gardens.

CEO of Cloudflare @eastdakota on agents as the ultimate infrastructure multiplier

@juliarturc The inevitable clawback

Oh no I haven’t yet tried OpenClaw and it’s already dead. Can I get an extension for a couple of weeks plz? 🥺

@dunkhippo33 • iTerm2 split terminal panes • worktrees, worktrees, worktrees • remote dev environments are next so all these environments don't step on each other for testing • Claude /rc when I leave my desk

Curious - in this world of constant building how people are setting up their computers and workflows? I have 4-5 terminals going at any given time across 2 Mac minis. right now it feels like I’m playing the piano across the two keyboards and terminals.  always looking for ideas - what is your workflow?

@mattshumer_ 4.7 does seem like a major regression for general business logic. You have to prompt it much more specifically. It's more like Codex that way.

@mattshumer_ I haven't even tried Opus 4.7 on frontend. It's tearing through every devops and cybersecurity task Opus 4.6 couldn't handle

Am I the only one having a good experience with Opus 4.7? I still vastly prefer Codex for most things but Opus is absolutely nailing every UI task I give it.