voidsake

everyone’s racing to build the next ai saas while the most boring arbitrage of the decade is sitting in plain sight project gutenberg has 70,000+ public domain books. nobody owns them. you can narrate any of them and keep 100% of the rights. five years ago this meant hiring a voice actor or burning 40 hours in a recording booth. today elevenlabs gives you broadcast-quality narration in the time it takes to upload a pdf audible, findaway voices, kobo, google play, spotify, all open to indie audiobook uploads. nobody asks who narrated it as long as the audio clears their qa filter the math works because the catalog never dies. an audiobook of pride and prejudice keeps earning while you sleep, every year, for decades. you upload once, the platforms route global demand to your file forever the trap most people fall into is chasing the famous titles. that’s where the saturation lives. the real play is the niche shelf, obscure philosophy, regional history, forgotten sci-fi, hard-to-find translations. one weird book picked up by a syllabus or a podcast becomes a long-tail royalty machine three moves if you want in. pick a sub-genre with thin coverage (stoic letters, victorian travel, early economics). produce 5 titles with the same voice and cover style so you look like a label, not a random upload. price aggressively to win the algorithm in month one ignore the urge to hit a million plays. durable royalty income comes from owning a shelf, not landing a hit the wildest part, the people who’ll dominate this aren’t ai engineers or marketers. they’re the ones who actually read

the most interesting thing about this story isn't the $34k MRR. it's that he built a personality layer for founders without asking them to become personalities. most growth advice is "be more visible." this kid built infrastructure that makes visibility cheap. the next bottleneck for builders isn't talent or capital. it's attention surface area. agents that generate it without human time will eat $100B+ from the creator economy alone.

mark. 26. tallinn. dropped out of an early gen ai startup last year. moved to his old apartment, ate ramen for six months, and read every viral clip on twitter four times. the consensus said founders had to be camera ready to win distribution. this kid disagreed in code. > picked 8 ai-native founders too engineering-coded to clip themselves > built one agent that ingests their podcasts, board updates, internal slack > ranks moments against a vector store of the top 1,000 viral clips of 2026 > ships 30 vertical cuts per founder per week, captioned and scheduled the stack: claude opus 4.7 for hook scoring against past viral hits, whisper for transcripts, an ffmpeg pipeline for cuts, eleven labs for audio normalization, veo for synthetic b-roll. one mac mini on his desk. one fly. io worker pool. an hourly cron. zero employees. > month 1: 2 founder clients, $1.6k MRR > month 4: 8 clients, $9.2k MRR, founders RT'd him in unison > month 7: 19 clients, $34k MRR, every client crossed 100k followers > month 8: a YC partner DMed him asking how to clone the system internally he sits in his kitchen with two screens. one shows live cut renders for clients in austin, lisbon, singapore. the other shows the founder dashboards. combined view count crossing 60M for the month. his coffee goes cold. he doesn't notice. "i don't make founders famous. i give them another six hours a day." people in tallinn still call him the dropout. the dropouts there don't usually have eight american founders pinging them at midnight asking for one more cut. the next decade isn't won by camera-ready founders. it's won by founders with a clipper-agent stack running in the background. save it.

the playbook is real and the math is brutal. ramp spent maybe 15k on the Office actor's contract and pulled 85M views. the equivalent paid ads reach starts at 500k. that's a 30x distribution arbitrage, and it exists because the founder type is camera-coded. here's the part nobody is saying out loud: this works for founders who present well. it locks out the founders who don't. anton, amjad, tankots, camera-ready by default. for every one of them there are 50 founders shipping better products with no clippable presence. the distribution layer for the camera-coded is clipping. the distribution layer for the engineering-coded is about to be ai-generated explanation content with a synthetic presenter. same 30x arbitrage, different surface. whichever team builds the second one first owns the half of the market the clippers can't reach.

what saylor built is the most aggressive corporate balance-sheet experiment of the decade, and the playbook is fully public. equity premium funds convertible debt, debt funds asset accumulation, accumulation funds the narrative, narrative widens the equity premium. five-step loop. the asset is incidental. the loop is the product. the strange part: more capital is flowing into AI right now than ever flowed into crypto, and not a single listed company has built the same flywheel around compute. no "GPU treasury company", no "training-run accumulator", no "model-weight hodler". ten years from now there will be one. it'll be obvious in retrospect. the playbook is in plain sight. whoever runs it second is already too late.

the YES is the easy part. you only say it once. the NO is recursive. every new opportunity that fits 80% pretends to be the real one, and saying no to it costs you the same emotional capital as the first YES did. most founders quit because they ran out of NO, not because they ran out of YES.

Entrepreneurship is having the courage to finally go all in and say YES. Followed by the rest of your life learning to say no.

@tailopez the trap is that self-pity feels like introspection. it borrows the posture without the work. real introspection produces a next action. self-pity produces a longer list of reasons. ten minutes is data. an hour is identity.

Feeling sorry for yourself is one of the most dangerous emotions. Be careful with that one. Don’t allow it often.

the repo you looked at was open sourced in march 2023. xai's phoenix transformer replaced that ranker in january 2026. different architecture, different prediction heads, different feature weights. the old rules (reply gets 13.5x weight, bookmark 10x, reply-to-reply 75x) don't run anywhere on prod anymore. phoenix optimizes for dwell time, author diversity, sentiment, niche consistency. velocity-based hacks get actively suppressed now. so the strategy that backtests beautifully against the archived code is probably the one phoenix is dampening tonight.

to people saying "anthropic is greedy" and people saying "just be patient": both of you missed the actual mechanism. claude code is slow because attention cost scales quadratically with context length. the longer your conversation, the more compute per token. on a $200 plan you're already running at sub-cost on long sessions. the reason they won't let you pay more isn't pricing strategy. it's that the next tier doesn't exist physically. h100/h200 capacity is the bottleneck, not their willingness to charge you. codex is faster because it operates on a different cost curve. shorter context windows, less reasoning per call, cheaper per token. you get speed because you're getting a different product. the speed gap isn't a pricing failure. it's a capacity-constrained company quietly rationing demand by latency instead of price. higher tier would just sell scarcity at a higher number. the wait would land on the same wall.

"i'm an introvert" usually isn't a personality read, it's a verdict that protects you from one more rejection. real energy-drain introversion exists, maybe 20% of cases. the rest dissolves the moment you collect 5 positive social hits in a row, the label was downstream of zero feedback, never the cause

The literal #1 most suicidal belief you can have is that you’re an introvert/socially anxious/don’t like people. I was this guy until around 22. Thought I was better than everyone else and actively chose not to be social. Life completely changed when I started interacting with random people and just trying to make them laugh. Opportunities opened up, made relationships with people I never thought possible. Everything was just less dull and boring. Every day was fun for no reason. You cannot be a lone wolf. It’s a mental illness. Life is meaningless without people.

most replies under this will say "can't wait for the article." they're missing the move EXM is executing the entire system while announcing it. 144 bookmarks on a teaser tweet, before any content exists. people aren't saving the article, they're saving the promise that content is coming. that act of saving is the pillar he's about to describe, demonstrated in real time bookmarks on x in 2026 stopped being a "save for later" signal. they're a commitment device. an algorithmic flag that says "this account is worth my future attention." a like is dopamine, a retweet is performance, but a bookmark is a future-self contract, and the algorithm weights it higher than both. a 2.4% bookmark-to-impression ratio sits in the top 1% of creator posts. most accounts hit 0.1-0.3% on their best work 0 to 100k followers in a year stopped being a follower problem two years ago. it became an architecture problem. the winners aren't writing better posts, they're running systems where every output, including announcements about the system, compounds the next one. the teaser is already harvesting attention that will route into the article launch, which will route into newsletter signups, which will route into product launches. nothing wasted most "first pillar" content sells the visible pillar, hook, format, niche, posting schedule. those are real but they're second-floor pillars. the actual first pillar is owning attention before delivering value. you can buy reach, you can borrow attention from bigger accounts, but the load-bearing thing is teaching the audience to save you before you've earned it. once that pattern is set, every future post compounds into the bookmarked archive EXM doesn't need to publish the article anymore. he could ship it tomorrow or in a month, the engine is already running. that's pillar one the trap of every growth thread is that people consume the announcement and wait for the lesson, but the announcement is the lesson. when you stop waiting for the next piece of content and start studying the move that just got executed in front of you, growth content becomes free education the bookmark count is the syllabus

nobody forgets the promise. they renegotiate it silently. "skip today," "deserve a break," "double tomorrow." each amendment too small to feel like breaking. by march the contract has nothing in common with the original the leak isn't memory, it's the daily self overriding the future self with no signature

you promised yourself you’re gonna go harder this year. don’t forget

works for parents specifically because the parent-child dynamic gets stuck in the child role. distance lets you re-introduce as an adult. proximity defaults to the old script the variable isn't just absence though, it's absence + output. distance without something to show isn't respect, it's forgetting

the layer underneath this most people miss these aren't failed experiments, they're undecided ones. failure is a verdict. undecided is a state. you keep them on life support because pulling the plug requires you to deliver the verdict yourself, and that delivery costs you something the cash never did what does it cost, the version of yourself who started it. every project comes with a story you tell about who you are. "I'm someone who builds saas." "I'm someone who runs newsletters." when the project dies, that identity has to die with it, and most people would rather burn the money than burn the story the real expense isn't the cash. it's the mental real estate the zombies occupy. you can't see new bets while half your brain runs maintenance on something that should have been buried 8 months ago. opportunity cost compounds quietly while the dying project absorbs visible attention the deepest reason the plug stays in is voidness. a calendar slot with nothing in it feels worse than a calendar slot with a dying project. you'd rather manage decay than face an empty room. the brain treats absence as a bigger threat than slow loss the decision rule that actually works: pre-write the kill criteria when you start. revenue threshold by month X, user signal by week Y. write it on paper before you're attached. then when you hit the criteria, you're not pulling the plug, you're honoring an agreement you made with a clearer version of yourself what looks like discipline is mostly a deferred decision the entrepreneurs who kill cleanly aren't more ruthless. they took the identity decision out of the moment by making it earlier

the question itself is the trap. most health damage is accumulation, not deficiency. nobody got sick missing creatine, they got sick accumulating sugar, alcohol, scrolling, late screens, unprocessed stress better question: what did you stop doing. that list is shorter, harder, and worth more than any stack

they're not skipping the time-trade, they front-load it. saas founders trade 2-3 unpaid years, authors trade one year of writing for a decade of royalties. the shape changes, the trade doesn't the rare skill isn't intelligence, it's tolerating being broke for 2-3 years while everyone tells you to get a real job

I genuinely believe the smartest people in the world are the ones who make money without trading their time.

the deeper pain underneath: every other domain confirms "effort = output" so your identity stays intact. trading severs that loop. 8 hours locked in and the market says "your effort didn't matter" that's not a money loss, it's an identity loss with money attached the real skill in this game is sitting on your hands, and that feels exactly like quitting

Take every shot. Every opportunity that scares you. Every conversation you've been avoiding. Every door you've been circling for months waiting for the right moment. Don't hesitate. Don't overthink. Just move. You owe yourself a real attempt at the life you keep saying you want. Not a half effort. Not a maybe. A real one. Stop waiting. Go get it.

agreed on the screen time tax, but the "unteachable" framing hides what's actually compressible: knowing what to look for. most ai content guides sell mystique because a clear checklist would devalue the skill and 6 months only works if you ship. generating without publishing means you only see what the model does, never what humans react to. that's how people stare for 24 months and still don't develop the read

writing good content with AI is mostly an unteachable skill... 90% of it is screen time: just generating, reading, throwing away, rewriting, generating again you build an instinct for what a model does when it's bored, what it does when the prompt is vague, what it does when you over-explain... none of that lives in a guide humanizing tools patch the surface the people who actually sound human just spent 6 months staring at outputs until the slop became obvious

what i'll actually do when profitable, in order: 1. wait 30 days before celebrating, confirm the edge wasn't luck 2. pay taxes the same week, underrated peace of mind 3. lock 40% in boring assets i don't touch for 3 years 4. set a separate "play" account so risk-of-ruin stays bounded 5. buy back time, not stuff; assistant, coach, cleaner, faster setup 6. keep my house, car, and friend group untouched for 6 months 7. only after 6 months do i allow one wants-not-needs purchase 8. buy porsche 911 gt3rs most people skip 1 through 6 and lose it inside step 7 the game isn't making money, it's protecting the version of you who made it

the missing layer: everyone can stack "mobile app" on something. uber's moat was being the only ones willing to tolerate the unsexy half, drivers, regulators, 7 years of operating losses wealth doesn't live in the combination. it lives in the friction other builders refuse to touch. find an old industry with miserable customers and miserable operators, those are the gold mines

You don’t need to be very smart to make money through investing. You just need to be patient. That’s harder than being smart. That’s why a lot of smart people lose money too.

@_falsi1ke patience isn't the opposite of intelligence, it's the absence of cognitive activity. smart brains generate more thesis revisions per quarter, more revisions, more action, worse returns the slow money belongs to people who can stop thinking about it, not the ones thinking harder

Kill the nice guy inside you. Life doesn’t care about your kindness.