rari

@0xwhrrari • 7,699 subscribers

building @kollectivexyz ai | prediction markets | solo setups dm always open... 444

Shorts

My Uber driver asked what I do for work. "Software." "Cool. Can you look at something?" He handed me his phone at a red light. Terminal. Claude chat. Green P&L. +$6,200. He drives Uber 4 days a week. Makes $1,100. Has a 2-year-old daughter. "Where did you find this?" "Your article. The 14,000 wallets one." He read it three months ago. Didn't understand half of it. Asked Claude to explain it like he's five. 214 messages. All during breaks between rides. Parked at gas stations. Waiting for pings. First thing Claude told him: 87% of wallets lose money. Don't be the 87%. He installed poly_data. Fed it to Claude. Found 47 wallets with Sharpe above 2.0. Filtered crypto only. Quarter Kelly. $200 starting bankroll. From his tips. Try copytrade bot here: 93 messages later Claude helped him build the 20-line brain from the article. Bayesian updates. EV filter at 5%. Fully automated. Last 45 days: → 480 trades → 91.3% win rate → +$6,200 Best trade: whale convergence on Fed rate cut. 4 wallets entered in 2 minutes. Entry $0.12. Resolved $1.00. +$1,760. While dropping off a passenger at JFK. The passenger tipped him $5. The bot made $1,760. His wife found the Telegram alerts on his phone. Thought he was texting another woman. He showed her the P&L curve. "Can you make me one?" "How long until you quit driving?" He looked at me through the rearview mirror. "I'm not stopping. Uber is my cover story." I wrote the article. He actually opened terminal.

My Uber driver asked what I do for work. "Software." "Cool. Can you look at something?" He handed me his phone at a red light. Terminal. Claude chat. Green P&L. +$6,200. He drives Uber 4 days a week. Makes $1,100. Has a 2-year-old daughter. "Where did you find this?" "Your article. The 14,000 wallets one." He read it three months ago. Didn't understand half of it. Asked Claude to explain it like he's five. 214 messages. All during breaks between rides. Parked at gas stations. Waiting for pings. First thing Claude told him: 87% of wallets lose money. Don't be the 87%. He installed poly_data. Fed it to Claude. Found 47 wallets with Sharpe above 2.0. Filtered crypto only. Quarter Kelly. $200 starting bankroll. From his tips. Try copytrade bot here: 93 messages later Claude helped him build the 20-line brain from the article. Bayesian updates. EV filter at 5%. Fully automated. Last 45 days: → 480 trades → 91.3% win rate → +$6,200 Best trade: whale convergence on Fed rate cut. 4 wallets entered in 2 minutes. Entry $0.12. Resolved $1.00. +$1,760. While dropping off a passenger at JFK. The passenger tipped him $5. The bot made $1,760. His wife found the Telegram alerts on his phone. Thought he was texting another woman. He showed her the P&L curve. "Can you make me one?" "How long until you quit driving?" He looked at me through the rearview mirror. "I'm not stopping. Uber is my cover story." I wrote the article. He actually opened terminal.

156,956 Aufrufe

My mass communications professor mass-emailed the wrong file to 400 students. It wasn't grades. It wasn't the syllabus. It was a spreadsheet. 2,300 Polymarket positions. Entries, exits, timestamps. Seven figures in volume. He recalled it in 9 minutes. I already had it saved. Opened it at 2am. Couldn't sleep. Every winning trade had the same pattern - he was buying contracts where the price was way off from what the data said. Not gut calls. Math. I spent 4 days trying to reverse-engineer his edge. Pulled 86M fills from one repo: Fed everything into Claude. "Compare true probability vs market price on every contract updated in the last 6 hours." First version was garbage. 52% hit rate. Worse than a coin flip. Then I added one filter - only enter when Claude's estimate diverges from market price by more than 14%. Everything changed. Scanner runs every 11 seconds. 8 bots splitting the work. Claude scores every contract before entry. Try to copytrade: $1,500 starting bankroll. 7 weeks later: → 191 trades → 74% win rate → +$8,340 Saw him in lecture last Thursday. He was reading something on his phone under the desk. It was my tweet about prediction markets. He looked up. We made eye contact. He didn't say anything. But he smiled. I think he knows.

My mass communications professor mass-emailed the wrong file to 400 students. It wasn't grades. It wasn't the syllabus. It was a spreadsheet. 2,300 Polymarket positions. Entries, exits, timestamps. Seven figures in volume. He recalled it in 9 minutes. I already had it saved. Opened it at 2am. Couldn't sleep. Every winning trade had the same pattern - he was buying contracts where the price was way off from what the data said. Not gut calls. Math. I spent 4 days trying to reverse-engineer his edge. Pulled 86M fills from one repo: Fed everything into Claude. "Compare true probability vs market price on every contract updated in the last 6 hours." First version was garbage. 52% hit rate. Worse than a coin flip. Then I added one filter - only enter when Claude's estimate diverges from market price by more than 14%. Everything changed. Scanner runs every 11 seconds. 8 bots splitting the work. Claude scores every contract before entry. Try to copytrade: $1,500 starting bankroll. 7 weeks later: → 191 trades → 74% win rate → +$8,340 Saw him in lecture last Thursday. He was reading something on his phone under the desk. It was my tweet about prediction markets. He looked up. We made eye contact. He didn't say anything. But he smiled. I think he knows.

10,908 Aufrufe

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Anthropic Research Lead: "99% of our engineers run swarms of 300+ self-improving agents" "Close the loop, give the model a way to verify its own output" In a 20-minute session, an Anthropic Team member breaks down how to build agents that improve themselves The real setup is Claude running through loops, plan mode, and dynamic workflows Better than most $300 agent courses Bookmark and watch the talk Then read the article below

Anthropic Research Lead: "99% of our engineers run swarms of 300+ self-improving agents" "Close the loop, give the model a way to verify its own output" In a 20-minute session, an Anthropic Team member breaks down how to build agents that improve themselves The real setup is Claude running through loops, plan mode, and dynamic workflows Better than most $300 agent courses Bookmark and watch the talk Then read the article below

378,669 Aufrufe • vor 18 Tagen

Google CEO, Sundar Pichai: "If you don't learn how to orchestrate agents now, you'll spend 2027 catching up to people who started today" In 30 minutes, he explains why the best engineers are moving from writing code to running agents One agent researches One writes One tests One reviews One fixes The human becomes the operator, not the bottleneck Bookmark and watch the interview

Google CEO, Sundar Pichai: "If you don't learn how to orchestrate agents now, you'll spend 2027 catching up to people who started today" In 30 minutes, he explains why the best engineers are moving from writing code to running agents One agent researches One writes One tests One reviews One fixes The human becomes the operator, not the bottleneck Bookmark and watch the interview

343,903 Aufrufe • vor 23 Tagen

Most people think prompting is just writing better sentences Anthropic shows why that's wrong This 32-minute workshop breaks down how prompts are built for production Not prompt hacks Not "act as" Not another $300 course Actual systems Evals Edge cases Model migration Agent loops Good prompt is not one message It's a workflow the model can follow, test and improve This matters because every new claude opus or sonnet will still reward the same rules Watch the prompting playbook Bookmark it before your next bad ai answer is just a bad setup

Most people think prompting is just writing better sentences Anthropic shows why that's wrong This 32-minute workshop breaks down how prompts are built for production Not prompt hacks Not "act as" Not another $300 course Actual systems Evals Edge cases Model migration Agent loops Good prompt is not one message It's a workflow the model can follow, test and improve This matters because every new claude opus or sonnet will still reward the same rules Watch the prompting playbook Bookmark it before your next bad ai answer is just a bad setup

390,180 Aufrufe • vor 1 Monat

NVIDIA CEO Jensen Huang: "Nobody is just writing prompts anymore, the new job is building and running loops" He says this is the shift that defines the rest of 2026 In 23 minutes, he explains why the best engineers are moving from prompting to looping Prompt Output Check Fix Run again That is the new workflow Bookmark and watch the interview Then save the full framework below

NVIDIA CEO Jensen Huang: "Nobody is just writing prompts anymore, the new job is building and running loops" He says this is the shift that defines the rest of 2026 In 23 minutes, he explains why the best engineers are moving from prompting to looping Prompt Output Check Fix Run again That is the new workflow Bookmark and watch the interview Then save the full framework below

169,352 Aufrufe • vor 21 Tagen

Ran into an Anthropic engineer at an airport lounge in Frankfurt last month. SFO delayed six hours. Nothing to do. I had my laptop open. Bot running. Three agents scoring markets in the background. Guy sits two seats down. Glances at my screen. Looks away. Looks back. "Is that Polymarket?" I said yeah. "You're running it with Claude Code?" I nodded. He pulled his chair closer. "I'm on the agent team. We stress-test this exact setup internally. You built it in your kitchen" I asked what stress-test meant. "Red team. We simulate what happens when someone gives Claude raw wallet data and tells it to find the winners. You're running the thing we were scared about" I showed him the scanner. One prompt. Find every wallet with 100 plus trades, 70 plus win rate. Rank by profit. Export top 50. Claude chewed through 14,000 wallets in under 4 minutes. Came back with 47. Top 20 made more than the bottom 13,000 combined. "That's not a distribution. That's a hit list" I said yeah. "And the scoring function?" Claude wrote it. I just wrapped it in an if-statement. A fill landed on screen mid-conversation. +$128 on an ETH dominance market. He stared at it. "How does it decide to enter" Three agents. Shared wallet. No shared memory. Arbitrage, convergence, whale copy. 2 of 3 agree, full size. 1 alone, half. Disagree, no trade. That consensus filter alone killed 40% of losing trades. "And the exit" The 47 whales never hold to settlement. 91% exit early. Capture 73% of max. I cut at 85% of the expected move. Or a 3x volume spike. Whichever comes first. "So you built a whale copy bot that exits before the whales" Yeah. He put his coffee down. "That's exactly the thing we gamed out in March. Management killed the internal version on policy. Nobody outside the team should have shipped this yet" I asked what he was going to do. "Fly home and pretend I didn't see it" $600 seed. 31 days. Net +$35,200. 479 trades. 75% win rate. Sharpe 2.63. I haven't touched the bot in three weeks. Copytrade for those who don't want to build - Before boarding he slid a card across the table. "When the next agent model drops, rerun your prompt. You'll see what I mean"

Ran into an Anthropic engineer at an airport lounge in Frankfurt last month. SFO delayed six hours. Nothing to do. I had my laptop open. Bot running. Three agents scoring markets in the background. Guy sits two seats down. Glances at my screen. Looks away. Looks back. "Is that Polymarket?" I said yeah. "You're running it with Claude Code?" I nodded. He pulled his chair closer. "I'm on the agent team. We stress-test this exact setup internally. You built it in your kitchen" I asked what stress-test meant. "Red team. We simulate what happens when someone gives Claude raw wallet data and tells it to find the winners. You're running the thing we were scared about" I showed him the scanner. One prompt. Find every wallet with 100 plus trades, 70 plus win rate. Rank by profit. Export top 50. Claude chewed through 14,000 wallets in under 4 minutes. Came back with 47. Top 20 made more than the bottom 13,000 combined. "That's not a distribution. That's a hit list" I said yeah. "And the scoring function?" Claude wrote it. I just wrapped it in an if-statement. A fill landed on screen mid-conversation. +$128 on an ETH dominance market. He stared at it. "How does it decide to enter" Three agents. Shared wallet. No shared memory. Arbitrage, convergence, whale copy. 2 of 3 agree, full size. 1 alone, half. Disagree, no trade. That consensus filter alone killed 40% of losing trades. "And the exit" The 47 whales never hold to settlement. 91% exit early. Capture 73% of max. I cut at 85% of the expected move. Or a 3x volume spike. Whichever comes first. "So you built a whale copy bot that exits before the whales" Yeah. He put his coffee down. "That's exactly the thing we gamed out in March. Management killed the internal version on policy. Nobody outside the team should have shipped this yet" I asked what he was going to do. "Fly home and pretend I didn't see it" $600 seed. 31 days. Net +$35,200. 479 trades. 75% win rate. Sharpe 2.63. I haven't touched the bot in three weeks. Copytrade for those who don't want to build - Before boarding he slid a card across the table. "When the next agent model drops, rerun your prompt. You'll see what I mean"

618,409 Aufrufe • vor 2 Monaten

Andrew Ng: "100% of my tasks now run through AI agents, the hype actually passed my expectations, loops are the next step" "In 3-6 months, everyone will be using self-improving loops, prompting alone is over" In a 30-minute talk, Andrew Ng breaks down how to build self-improving agentic systems from scratch Agent loops Feedback Memory Self-improvement Worth more than most $500 agentic courses

Andrew Ng: "100% of my tasks now run through AI agents, the hype actually passed my expectations, loops are the next step" "In 3-6 months, everyone will be using self-improving loops, prompting alone is over" In a 30-minute talk, Andrew Ng breaks down how to build self-improving agentic systems from scratch Agent loops Feedback Memory Self-improvement Worth more than most $500 agentic courses

66,294 Aufrufe • vor 16 Tagen

Anthropic CEO Dario Amodei: "We didn't plan for 80x growth, it would not have been rational" In 47 minutes, he explains why Anthropic is sitting on Mythos even though it could print money tomorrow Early companies called it a super weapon One said you should need a gun license to use it Mythos already found 271 new firefox vulnerabilities Plus thousands more in private codebases The model is so useful they are losing money by not releasing it Bookmark and watch the interview

Anthropic CEO Dario Amodei: "We didn't plan for 80x growth, it would not have been rational" In 47 minutes, he explains why Anthropic is sitting on Mythos even though it could print money tomorrow Early companies called it a super weapon One said you should need a gun license to use it Mythos already found 271 new firefox vulnerabilities Plus thousands more in private codebases The model is so useful they are losing money by not releasing it Bookmark and watch the interview

81,463 Aufrufe • vor 22 Tagen

I gave Claude one Polymarket equation and told it to find every mispriced contract It hasn't stopped running in 4 months $600 → $14,190 Every Polymarket contract is priced by one formula: C(q) = b × ln(Σ e^(qi / b)) The pricing function inside it is softmax - the same math Claude uses to pick its next token Claude doesn't just understand this equation, it thinks in it So I gave it a task: "Scan every open contract, compute fair value, flag anything the market has wrong 24/7" It finds the gap, sizes the position by Kelly, enters before the crowd corrects Market shows 0.38, math shows 0.57, bot buys Resolution hits, bot collects 120 days 61.2% win rate 2.31 sharpe -3.8% max drawdown 425 trades 87% of Polymarket wallets are underwater They trade feelings against a formula that doesn't bluff, doesn't tilt, doesn't chase losses The edge is having something that reads the math faster than any human - and never stops looking

I gave Claude one Polymarket equation and told it to find every mispriced contract It hasn't stopped running in 4 months $600 → $14,190 Every Polymarket contract is priced by one formula: C(q) = b × ln(Σ e^(qi / b)) The pricing function inside it is softmax - the same math Claude uses to pick its next token Claude doesn't just understand this equation, it thinks in it So I gave it a task: "Scan every open contract, compute fair value, flag anything the market has wrong 24/7" It finds the gap, sizes the position by Kelly, enters before the crowd corrects Market shows 0.38, math shows 0.57, bot buys Resolution hits, bot collects 120 days 61.2% win rate 2.31 sharpe -3.8% max drawdown 425 trades 87% of Polymarket wallets are underwater They trade feelings against a formula that doesn't bluff, doesn't tilt, doesn't chase losses The edge is having something that reads the math faster than any human - and never stops looking

311,727 Aufrufe • vor 3 Monaten

Got laid off on a Friday Spent the weekend reading Polymarket docs instead of updating my resume By Monday I had 3 agents running on a Mac Mini $500 → $3,140 Before my boss even posted my replacement job listing Weather data is public NOAA publishes forecasts every 6 hours The models are accurate But the market prices contracts off weather apps and vibes That's not a gambling problem That's a latency problem So I built 3 agents to exploit it Agent - 01: Reads raw NOAA forecast grids Not weather The actual meteorological data Agent - 02: Compares NOAA confidence intervals against live Polymarket prices Finds the gap before it closes Agent - 03: Enters the position EIP-712 signature, direct to CLOB on Polygon No middleman, no delay Not because the agents are smart Because the crowd updates slowly And the agents update constantly 71.2% win rate $26.40 average 3.9% max drawdown Built a live terminal to watch it all run in real time Balance, P&L, win rate, active markets - all updating live And the agents report back directly: Agent - 01: Batch filled 12 positions Avg +$26 Edge holding Agent - 02: Cold front repricing 8 markets Entering now Agent - 03: 3 losses in a row Drawdown at 6.8% Need wider capital base to hold variance It's not glamorous It's just a black screen with green text And numbers going up Copytrading open:

Got laid off on a Friday Spent the weekend reading Polymarket docs instead of updating my resume By Monday I had 3 agents running on a Mac Mini $500 → $3,140 Before my boss even posted my replacement job listing Weather data is public NOAA publishes forecasts every 6 hours The models are accurate But the market prices contracts off weather apps and vibes That's not a gambling problem That's a latency problem So I built 3 agents to exploit it Agent - 01: Reads raw NOAA forecast grids Not weather The actual meteorological data Agent - 02: Compares NOAA confidence intervals against live Polymarket prices Finds the gap before it closes Agent - 03: Enters the position EIP-712 signature, direct to CLOB on Polygon No middleman, no delay Not because the agents are smart Because the crowd updates slowly And the agents update constantly 71.2% win rate $26.40 average 3.9% max drawdown Built a live terminal to watch it all run in real time Balance, P&L, win rate, active markets - all updating live And the agents report back directly: Agent - 01: Batch filled 12 positions Avg +$26 Edge holding Agent - 02: Cold front repricing 8 markets Entering now Agent - 03: 3 losses in a row Drawdown at 6.8% Need wider capital base to hold variance It's not glamorous It's just a black screen with green text And numbers going up Copytrading open:

324,995 Aufrufe • vor 4 Monaten

Anthropic CEO Dario Amodei: "Some early companies got Mythos and told us this is a super weapon, please don't release it" On leaving OpenAI: "When you can't trust someone, and their values don't match what they say, it becomes very hard to stay" Boris Cherny, head of Claude Code: "90% of code at Anthropic is written by Claude, a lot of Claude Code is written by Claude Code" Then Dario says the quiet part even louder: 10–25% odds of civilizational collapse from AI 47 minutes Free Dario, Daniela, Boris Probably the deepest look inside Anthropic ever filmed Mythos Claude Code OpenAI split Superweapon models 90% AI-written code Bookmark it and watch ↓

Anthropic CEO Dario Amodei: "Some early companies got Mythos and told us this is a super weapon, please don't release it" On leaving OpenAI: "When you can't trust someone, and their values don't match what they say, it becomes very hard to stay" Boris Cherny, head of Claude Code: "90% of code at Anthropic is written by Claude, a lot of Claude Code is written by Claude Code" Then Dario says the quiet part even louder: 10–25% odds of civilizational collapse from AI 47 minutes Free Dario, Daniela, Boris Probably the deepest look inside Anthropic ever filmed Mythos Claude Code OpenAI split Superweapon models 90% AI-written code Bookmark it and watch ↓

65,014 Aufrufe • vor 29 Tagen

OpenAI just mass-fired their robotics team. One of the engineers DM'd me 20 minutes later. I didn't know him. He found me through a Polymarket thread. His first message: "I have 30 days of severance and nothing to lose. Let me tell you what we actually use internally. It's not GPT" I thought he was trolling. "Every serious team at OpenAI prototypes on Claude Code. Not ChatGPT. Not the API. Claude Code connected to a repo. That's the actual workflow" I asked why. "Because Claude reads the codebase. GPT reads the prompt. There's a difference. One guesses. The other one understands the full context and builds on top of it" He sent me one link. 86 million Polymarket trades. Every wallet. Every entry. Every exit. Open source. Free. "Point Claude Code at this. Say - find every wallet with 70%+ win rate and 100+ trades. Watch what happens" I did it that night. Claude pulled 47 wallets in 4 minutes. Average profit: $214K. Hold time: 7 hours. 91% exit BEFORE resolution. They never wait for the outcome. "Now look at how they exit" Top wallets capture 86% of the move and cut at 12%. Everyone else captures 58% and holds losers to 41%. Same entries. Completely different results. He sent another link. "Three commands. Your bot sees 500+ markets. No key needed. Read-only. Claude scores them in 20 minutes" I asked why he's telling me all this. "Because I just got fired for saying we should open-source more. So here I am open-sourcing everything I know" Then he sent me an article where someone built the full bot from these repos in a weekend -> Three exit triggers: Target 85% of move. Volume spike x3 - smart money out. 24h silence - thesis dead. I copied the stack. Claude Code $20. VPS $5. $25/month. No team. No office. No GPT subscription. 17 days. 191 trades. 73% win rate. $850 seed. +$9,400. I sent him my results. He replied: "This is exactly what I built as a side project at OpenAI. They made me delete it" I asked if I could post this. "Post it. What are they gonna do. Fire me again?"

OpenAI just mass-fired their robotics team. One of the engineers DM'd me 20 minutes later. I didn't know him. He found me through a Polymarket thread. His first message: "I have 30 days of severance and nothing to lose. Let me tell you what we actually use internally. It's not GPT" I thought he was trolling. "Every serious team at OpenAI prototypes on Claude Code. Not ChatGPT. Not the API. Claude Code connected to a repo. That's the actual workflow" I asked why. "Because Claude reads the codebase. GPT reads the prompt. There's a difference. One guesses. The other one understands the full context and builds on top of it" He sent me one link. 86 million Polymarket trades. Every wallet. Every entry. Every exit. Open source. Free. "Point Claude Code at this. Say - find every wallet with 70%+ win rate and 100+ trades. Watch what happens" I did it that night. Claude pulled 47 wallets in 4 minutes. Average profit: $214K. Hold time: 7 hours. 91% exit BEFORE resolution. They never wait for the outcome. "Now look at how they exit" Top wallets capture 86% of the move and cut at 12%. Everyone else captures 58% and holds losers to 41%. Same entries. Completely different results. He sent another link. "Three commands. Your bot sees 500+ markets. No key needed. Read-only. Claude scores them in 20 minutes" I asked why he's telling me all this. "Because I just got fired for saying we should open-source more. So here I am open-sourcing everything I know" Then he sent me an article where someone built the full bot from these repos in a weekend -> Three exit triggers: Target 85% of move. Volume spike x3 - smart money out. 24h silence - thesis dead. I copied the stack. Claude Code $20. VPS $5. $25/month. No team. No office. No GPT subscription. 17 days. 191 trades. 73% win rate. $850 seed. +$9,400. I sent him my results. He replied: "This is exactly what I built as a side project at OpenAI. They made me delete it" I asked if I could post this. "Post it. What are they gonna do. Fire me again?"

157,384 Aufrufe • vor 2 Monaten

This girl make $10,000/month from kids videos All from a phone Find a viral abc song on YouTube Copy the lyrics from the description Ask ChatGPT for a cartoon prompt Paste it into an AI video tool. Puppies, letters, balloons, bright colors. No face, no camera, no team Upload daily Kids replay this stuff nonstop YouTube pays for the loop Bookmark this before everyone tries it

This girl make $10,000/month from kids videos All from a phone Find a viral abc song on YouTube Copy the lyrics from the description Ask ChatGPT for a cartoon prompt Paste it into an AI video tool. Puppies, letters, balloons, bright colors. No face, no camera, no team Upload daily Kids replay this stuff nonstop YouTube pays for the loop Bookmark this before everyone tries it

80,493 Aufrufe • vor 1 Monat

My girlfriend said "you've been staring at that Polymarket screen for 6 hours" I turned the laptop around She went quiet That screen had $4,100 more than when I started $150 → $38,700 in 31 days 4 agents running on a $400 laptop, no breaks, no sleep: Agent-01 WEATHER: Pulls NOAA forecast grids every 10 minutes Finds cities where federal models say 92% but Polymarket says 50% Buys at 8¢, exits at 50¢+ Agent-02 BTC 5min/15min: Tracks Binance spread vs Polymarket CLOB Enters when the gap hits 3%+ Closes before resolution Agent-03 POLITICS: Scrapes polling data and sentiment shifts Flags contracts lagging behind real movement Enters before the crowd reprices Agent-04 SPORTS: Reads injury reports and line movement Finds mispriced live markets Executes via EIP-712 on Polygon The edge is embarrassingly simple NOAA has a $6.5 billion supercomputer Retail prices weather contracts off vibes and weather apps That's not a fair fight That's federal science vs guessing Chicago - NOAA says 92% chance of rain Polymarket contract sitting at 8¢ 6x return on public data 28 trades across 6 cities in one night 2,900+ trades 91% win rate $38,700 from $150 Agent-01: NOAA grid update received. 3 new mispricings detected. executing... Agent-02: BTC spread widened to 4.1%. entering 15min YES at 42¢ Agent-03: sentiment shift on midterm contract. buying NO at 11¢ Agent-04: injury report confirmed. line moving. entering at 23¢ She still doesn't understand what's on the screen She doesn't need to

My girlfriend said "you've been staring at that Polymarket screen for 6 hours" I turned the laptop around She went quiet That screen had $4,100 more than when I started $150 → $38,700 in 31 days 4 agents running on a $400 laptop, no breaks, no sleep: Agent-01 WEATHER: Pulls NOAA forecast grids every 10 minutes Finds cities where federal models say 92% but Polymarket says 50% Buys at 8¢, exits at 50¢+ Agent-02 BTC 5min/15min: Tracks Binance spread vs Polymarket CLOB Enters when the gap hits 3%+ Closes before resolution Agent-03 POLITICS: Scrapes polling data and sentiment shifts Flags contracts lagging behind real movement Enters before the crowd reprices Agent-04 SPORTS: Reads injury reports and line movement Finds mispriced live markets Executes via EIP-712 on Polygon The edge is embarrassingly simple NOAA has a $6.5 billion supercomputer Retail prices weather contracts off vibes and weather apps That's not a fair fight That's federal science vs guessing Chicago - NOAA says 92% chance of rain Polymarket contract sitting at 8¢ 6x return on public data 28 trades across 6 cities in one night 2,900+ trades 91% win rate $38,700 from $150 Agent-01: NOAA grid update received. 3 new mispricings detected. executing... Agent-02: BTC spread widened to 4.1%. entering 15min YES at 42¢ Agent-03: sentiment shift on midterm contract. buying NO at 11¢ Agent-04: injury report confirmed. line moving. entering at 23¢ She still doesn't understand what's on the screen She doesn't need to

166,409 Aufrufe • vor 4 Monaten

Anthropic's managed agents team just showed the Fable 5 workflow most people haven't seen Their line was simple: Fable 5 is our best model for running self-improving agent systems" Fable 5 + loops + dynamic workflows + dreaming = self-improving agents 13 minutes Live from Anthropic's Japan stage Built from scratch Worth more than a $500 agent building course Bookmark and watch it before this becomes the default setup

Anthropic's managed agents team just showed the Fable 5 workflow most people haven't seen Their line was simple: Fable 5 is our best model for running self-improving agent systems" Fable 5 + loops + dynamic workflows + dreaming = self-improving agents 13 minutes Live from Anthropic's Japan stage Built from scratch Worth more than a $500 agent building course Bookmark and watch it before this becomes the default setup

32,231 Aufrufe • vor 27 Tagen

Claude Opus 4.8 should be the final reason to stop writing specs by hand Anthropic's Claude Code architect shows the new workflow in 31 minutes The model interviews you first Then it pulls out the requirements better than your own 200-line markdown doc No dead specs No messy notes No file nobody reads Claude writes the spec Renders it in html Checks the result inside the same file Opus 4.8 Auto mode X high effort This is less like prompting More like letting the agent understand the job before it builds Bookmark and watch the video below

Claude Opus 4.8 should be the final reason to stop writing specs by hand Anthropic's Claude Code architect shows the new workflow in 31 minutes The model interviews you first Then it pulls out the requirements better than your own 200-line markdown doc No dead specs No messy notes No file nobody reads Claude writes the spec Renders it in html Checks the result inside the same file Opus 4.8 Auto mode X high effort This is less like prompting More like letting the agent understand the job before it builds Bookmark and watch the video below

39,164 Aufrufe • vor 1 Monat

An Anthropic engineer left his laptop unlocked at a cafe in SF I wasn't trying to look. But the screen was right there. A Polymarket terminal. Live trades scrolling. Green numbers. And a model ID I'd never seen before - claude-opus-4-7. He came back. Saw me staring. Didn't close it. "You trade?" I told him I run 8 Claude agents on Polymarket. He sat down like I just said a password. "What model are you on?" 4.6. He shook his head. "We've been running 4.7 internally for two weeks. Your disposition filter - what is it?" 0.70. He laughed. "Ours is 0.82. Below that you're literally copying noise. But on 4.7 you can push it to 0.58. The model filters its own noise now" He turned his laptop. A spreadsheet. 47 wallets ranked by exit quality. "Not win rate. Exit quality. Top wallets capture 86% of the move and cut at 12%. Everyone else captures 58% and holds losers past 40%" I've never heard anyone break it down like that. He opened one repo. 86 million trades. Every wallet. Every entry. Every exit. Public. Then a second. "Three commands. Every order book. No API key. 4.7 scores them in seconds. Not minutes. Seconds" I asked why he's telling me this. "Because my side bot makes more than my Anthropic salary and I want the liquidity to stay deep" He showed me his phone. +$47,000. 38 days On a model that wasn't even public yet. I asked when 4.7 drops. He said "this week". It dropped today. I switched all 8 agents in 10 minutes. Rebuilt everything that night. First week: +$2,800 Second: +$2,400 Third: +$2,100 Fourth: +$2,100 +$9,400 in 28 days. 191 trades. 73% win rate. Avg hold 4h. Crypto +$3,740 Weather +$2,910 Politics +$1,850 Macro +$1,200 Sports -$300 Killed it after day 2 Capital velocity: 47x. Kelly f+ 0.068. Max drawdown -2.1%. Bot for those who don't build: I asked if I could post this. "Don't use my name and we're good" Done.

An Anthropic engineer left his laptop unlocked at a cafe in SF I wasn't trying to look. But the screen was right there. A Polymarket terminal. Live trades scrolling. Green numbers. And a model ID I'd never seen before - claude-opus-4-7. He came back. Saw me staring. Didn't close it. "You trade?" I told him I run 8 Claude agents on Polymarket. He sat down like I just said a password. "What model are you on?" 4.6. He shook his head. "We've been running 4.7 internally for two weeks. Your disposition filter - what is it?" 0.70. He laughed. "Ours is 0.82. Below that you're literally copying noise. But on 4.7 you can push it to 0.58. The model filters its own noise now" He turned his laptop. A spreadsheet. 47 wallets ranked by exit quality. "Not win rate. Exit quality. Top wallets capture 86% of the move and cut at 12%. Everyone else captures 58% and holds losers past 40%" I've never heard anyone break it down like that. He opened one repo. 86 million trades. Every wallet. Every entry. Every exit. Public. Then a second. "Three commands. Every order book. No API key. 4.7 scores them in seconds. Not minutes. Seconds" I asked why he's telling me this. "Because my side bot makes more than my Anthropic salary and I want the liquidity to stay deep" He showed me his phone. +$47,000. 38 days On a model that wasn't even public yet. I asked when 4.7 drops. He said "this week". It dropped today. I switched all 8 agents in 10 minutes. Rebuilt everything that night. First week: +$2,800 Second: +$2,400 Third: +$2,100 Fourth: +$2,100 +$9,400 in 28 days. 191 trades. 73% win rate. Avg hold 4h. Crypto +$3,740 Weather +$2,910 Politics +$1,850 Macro +$1,200 Sports -$300 Killed it after day 2 Capital velocity: 47x. Kelly f+ 0.068. Max drawdown -2.1%. Bot for those who don't build: I asked if I could post this. "Don't use my name and we're good" Done.

71,528 Aufrufe • vor 2 Monaten

Ran into a Chinese ML researcher at a rooftop bar in Hudson Yards on Saturday. DeepMind lanyard in his pocket. Two glasses of baijiu in. Loose tie. Laptop open on the bar. Position check running. He saw the screen. "Is that Polymarket" Yeah. "You trade it with Opus 4.7" I nodded. He dragged a stool over. "I ran the internal eval on 4.7 last month. Best reasoning model we've tested on adversarial time-series. Nobody outside the lab is using it for money yet. You're the first I've seen" I asked what the eval looked like. "We hand the model raw wallet data and tell it to find signal. 4.7 finds it faster than our quants. Then legal kills the experiment. Same every quarter" I showed him the scanner. 86M trades. Public dataset nobody outside maintainers touches it. 14,000 wallets in. Filter by floor WR, trade count, size, recency. 9 wallets out. Opus 4.7 chewed it in 38 minutes on a $5 VPS. "Why 4.7 and not Codex" Codex writes code. 4.7 reads behavior. Different problem. "The filter" 40-200 trades. 74+ lifetime WR. Sub $1,500 avg. Active last 45 days. 3 qualifying 21-day windows. "Why those bands" Above 200 is a fund. Below 40 is luck. Between is the edge. A fill landed mid-conversation. +$128 on BTC+100K 5-min. "How does it enter" Each wallet only prints in its zone. 83% crypto. 51% politics. 49% sports. Copy only inside the zone. "Domain conditioning. We published on it in March. Nobody ships it retail" I ship it. Four Python scripts. Claude Code wrote 80% in a weekend. Delay 21 min. Bet size is the signal. $900 normal, $1,600 is 1.8 sigma. Size up. Exit mirrors source. 74.1% of max gain. Stale past 18h under 3% gets cut. Sharpe 2.67 453 trades 81.7% WR $600 seed +$41,985 in 31 days He set the glass down. "4.7 is a weapon on this data. Most people use it to write essays. You built a printer" He leaned in. "One more thing. Rerun the filter with 4.7 every 7 days. The wallet set rotates. Your 9 today are not your 9 in a month" Execution orderbook layer, no middleman. For those who don't build - He pulled out his phone, typed something, stopped. "If anyone at the lab asks, we never spoke" I said we didn't. I found 4.7 myself. He smiled. "Good answer" He finished the baijiu and walked to the elevator.

Ran into a Chinese ML researcher at a rooftop bar in Hudson Yards on Saturday. DeepMind lanyard in his pocket. Two glasses of baijiu in. Loose tie. Laptop open on the bar. Position check running. He saw the screen. "Is that Polymarket" Yeah. "You trade it with Opus 4.7" I nodded. He dragged a stool over. "I ran the internal eval on 4.7 last month. Best reasoning model we've tested on adversarial time-series. Nobody outside the lab is using it for money yet. You're the first I've seen" I asked what the eval looked like. "We hand the model raw wallet data and tell it to find signal. 4.7 finds it faster than our quants. Then legal kills the experiment. Same every quarter" I showed him the scanner. 86M trades. Public dataset nobody outside maintainers touches it. 14,000 wallets in. Filter by floor WR, trade count, size, recency. 9 wallets out. Opus 4.7 chewed it in 38 minutes on a $5 VPS. "Why 4.7 and not Codex" Codex writes code. 4.7 reads behavior. Different problem. "The filter" 40-200 trades. 74+ lifetime WR. Sub $1,500 avg. Active last 45 days. 3 qualifying 21-day windows. "Why those bands" Above 200 is a fund. Below 40 is luck. Between is the edge. A fill landed mid-conversation. +$128 on BTC+100K 5-min. "How does it enter" Each wallet only prints in its zone. 83% crypto. 51% politics. 49% sports. Copy only inside the zone. "Domain conditioning. We published on it in March. Nobody ships it retail" I ship it. Four Python scripts. Claude Code wrote 80% in a weekend. Delay 21 min. Bet size is the signal. $900 normal, $1,600 is 1.8 sigma. Size up. Exit mirrors source. 74.1% of max gain. Stale past 18h under 3% gets cut. Sharpe 2.67 453 trades 81.7% WR $600 seed +$41,985 in 31 days He set the glass down. "4.7 is a weapon on this data. Most people use it to write essays. You built a printer" He leaned in. "One more thing. Rerun the filter with 4.7 every 7 days. The wallet set rotates. Your 9 today are not your 9 in a month" Execution orderbook layer, no middleman. For those who don't build - He pulled out his phone, typed something, stopped. "If anyone at the lab asks, we never spoke" I said we didn't. I found 4.7 myself. He smiled. "Good answer" He finished the baijiu and walked to the elevator.

37,516 Aufrufe • vor 2 Monaten

Ran into a Chinese quant at a noodle spot in Flushing last week. Jane Street badge on his belt. Came straight from the PATH. Laptop open on the counter. Position check running. He stopped. "Is that Polymarket" Yeah. "You trade it with Claude" I nodded. He sat down without asking. "I'm on the systematic desk. We model this venue in research. You're running it from a noodle shop" I asked what the desk does with it. "We don't trade. Compliance won't let us. We run shadow books. Cleanest retail venue on earth. Every wallet public. Every fill on-chain. You're sitting on a dataset we'd pay seven figures for" I showed him the scanner. 86M trades. - nobody outside maintainers touches it. 14,000 wallets in. Filter by floor WR, trade count, size, recency. 9 wallets out. Opus 4.7 chewed it in 38 minutes on a $5 VPS. "The filter" 40-200 trades. 74+ lifetime WR. Sub $1,500 avg. Active last 45 days. 3 qualifying 21-day windows. "Why those bands" Above 200 is a fund. Below 40 is luck. Between is the edge. A fill landed. +$112 on BTC+100K 5-min. "How does it enter" Each wallet only prints in its zone. 83% crypto. 51% politics. 49% sports. Copy only inside the zone. "Domain conditioning. We have a paper on it. Nobody ships it retail" I ship it. Four Python scripts. Claude Code wrote 80% in a weekend. Delay 21 min. Bet size is the signal. $900 normal, $1,600 is 1.8 sigma. Size up. Exit mirrors source. 74.1% of max gain. Stale past 18h under 3% gets cut. Sharpe 2.67 253 trades 72.7% WR $200 seed +$14,985 in 31 days He put his chopsticks down. "Add one thing. When 8+ wallets dump the same side of a 5-min BTC window in under 60s, fade them. Crowd is the signal" Execution For those who don't build - He slid a napkin across before leaving. "If anyone at my firm asks, we never spoke" I said we didn't. I found it myself. He smiled. "Good answer" Napkin said: "the crowd is the liquidity"

Ran into a Chinese quant at a noodle spot in Flushing last week. Jane Street badge on his belt. Came straight from the PATH. Laptop open on the counter. Position check running. He stopped. "Is that Polymarket" Yeah. "You trade it with Claude" I nodded. He sat down without asking. "I'm on the systematic desk. We model this venue in research. You're running it from a noodle shop" I asked what the desk does with it. "We don't trade. Compliance won't let us. We run shadow books. Cleanest retail venue on earth. Every wallet public. Every fill on-chain. You're sitting on a dataset we'd pay seven figures for" I showed him the scanner. 86M trades. - nobody outside maintainers touches it. 14,000 wallets in. Filter by floor WR, trade count, size, recency. 9 wallets out. Opus 4.7 chewed it in 38 minutes on a $5 VPS. "The filter" 40-200 trades. 74+ lifetime WR. Sub $1,500 avg. Active last 45 days. 3 qualifying 21-day windows. "Why those bands" Above 200 is a fund. Below 40 is luck. Between is the edge. A fill landed. +$112 on BTC+100K 5-min. "How does it enter" Each wallet only prints in its zone. 83% crypto. 51% politics. 49% sports. Copy only inside the zone. "Domain conditioning. We have a paper on it. Nobody ships it retail" I ship it. Four Python scripts. Claude Code wrote 80% in a weekend. Delay 21 min. Bet size is the signal. $900 normal, $1,600 is 1.8 sigma. Size up. Exit mirrors source. 74.1% of max gain. Stale past 18h under 3% gets cut. Sharpe 2.67 253 trades 72.7% WR $200 seed +$14,985 in 31 days He put his chopsticks down. "Add one thing. When 8+ wallets dump the same side of a 5-min BTC window in under 60s, fade them. Crowd is the signal" Execution For those who don't build - He slid a napkin across before leaving. "If anyone at my firm asks, we never spoke" I said we didn't. I found it myself. He smiled. "Good answer" Napkin said: "the crowd is the liquidity"

25,246 Aufrufe • vor 2 Monaten

I MADE 3 AI AGENTS FIGHT FOR MY MONEY ON POLYMARKET Claude: $100 → $10,264 in 24 hours OpenClaw: broke even OpenAI: liquidated Same market. Same bankroll. Same window. Three different outcomes. OpenAI got a prompt and an API key. Traded on vibes. Vibes cost $100. OpenClaw had Kelly sizing and filters. Stayed alive. Didn't print. Claude had Perplexity behind it. Deep Research across 47+ sources before a single trade fired. NOAA forecasts, top trader wallets, fee structures, forecast accuracy - full strategy map in 10 minutes. NOAA says 94% on a temperature bucket. Market prices it at 11¢. Claude buys. Corrects to 47¢. Sells. 6 cities. 60+ markets. 720 scans per day. Try to copytrady claudebot: Claude - 312 trades, 71% win rate, P&L: +$10,164 OpenClaw - 89 trades, 52% win rate, P&L: +$12 OpenAI - 41 trades, 34% win rate, P&L: -$100 Perplexity does the thinking. Claude pulls the trigger. Polymarket pays.

I MADE 3 AI AGENTS FIGHT FOR MY MONEY ON POLYMARKET Claude: $100 → $10,264 in 24 hours OpenClaw: broke even OpenAI: liquidated Same market. Same bankroll. Same window. Three different outcomes. OpenAI got a prompt and an API key. Traded on vibes. Vibes cost $100. OpenClaw had Kelly sizing and filters. Stayed alive. Didn't print. Claude had Perplexity behind it. Deep Research across 47+ sources before a single trade fired. NOAA forecasts, top trader wallets, fee structures, forecast accuracy - full strategy map in 10 minutes. NOAA says 94% on a temperature bucket. Market prices it at 11¢. Claude buys. Corrects to 47¢. Sells. 6 cities. 60+ markets. 720 scans per day. Try to copytrady claudebot: Claude - 312 trades, 71% win rate, P&L: +$10,164 OpenClaw - 89 trades, 52% win rate, P&L: +$12 OpenAI - 41 trades, 34% win rate, P&L: -$100 Perplexity does the thinking. Claude pulls the trigger. Polymarket pays.

28,040 Aufrufe • vor 3 Monaten

20 free Tembo runs/day sounds tiny until you actually do the math The stack: Opus 4.7, GPT 5.5, GLM 5.1. That means the "free tier" is not toy-model autocomplete It is frontier coding models doing real work 1 Bugfix - 1-3 runs 1 Refactor - 3-5 runs 1 PR review - 1 run 1 Architecture pass - 2-4 runs So 20 runs/day is enough for: 5-7 Focused bug fixes 3-4 Serious refactors 10+ PR reviews Full solo dev session Every day, for $0 That is $240-$1,200/year saved if it replaces a $20-$100/mo coding agent subscription Meanwhile most devs are still paying $20-$100/mo because they think the wrapper is the product It is not The product is output And if 20 free runs can ship the same PR, the pricing math gets very uncomfortable Bookmark it before everyone starts pretending they found this first

20 free Tembo runs/day sounds tiny until you actually do the math The stack: Opus 4.7, GPT 5.5, GLM 5.1. That means the "free tier" is not toy-model autocomplete It is frontier coding models doing real work 1 Bugfix - 1-3 runs 1 Refactor - 3-5 runs 1 PR review - 1 run 1 Architecture pass - 2-4 runs So 20 runs/day is enough for: 5-7 Focused bug fixes 3-4 Serious refactors 10+ PR reviews Full solo dev session Every day, for $0 That is $240-$1,200/year saved if it replaces a $20-$100/mo coding agent subscription Meanwhile most devs are still paying $20-$100/mo because they think the wrapper is the product It is not The product is output And if 20 free runs can ship the same PR, the pricing math gets very uncomfortable Bookmark it before everyone starts pretending they found this first

14,193 Aufrufe • vor 1 Monat