Nathan Stevenson

@niccruzpatane Even if it's going to be the largest IPO in human history, it's still underestimated.

@cohere optimizing for security, throughout and TTFT latency. This commitment to open-source for such a high quality model is exciting. Can't wait to run it through our evals!

HTML is the new Markdown for AI output - For busy business executives, entrepreneurs and my @HarvardAlumni getting started with @claudeai #cowork Thariq @trq212 sets the gold standard for clarity in communicating emerging technical methods. There is an art to the subtle turns of phrase, in his prompts that is fresh, exciting and vital. This one article will 3-5X the speed and quality of your work product. #AIforeveryone #Prompting #learned

100% AI Native starts with the data foundation + distributed infra. Integrate the enterprise stack. secure autonomous systems. Structure for agentic AI and tokenomics.

@okkhary @Swatch @SwatchUS @AudemarsPiguet I stayed to see what would happen and ended up spending a lovely hour at the IWC Shafhausen store. Now those are real watches !

@NathanStvnsn @Swatch @SwatchUS @AudemarsPiguet Or.... You could have also just turned around and went home and got it at a later time. Can't blame them for your impatience. They're gonna be available for about 8 months.

@craig_krull @Swatch It may as well be a turd now 💩

@Swatch Woohoo , I got two

@iamliamsheridan benchmark2

claude opus 4.7 read every cold email we sent in q1. 4.7 million sends across 11 client accounts. a few things that surprised us: → tuesday is no longer the best day to send (wednesday narrowly wins) → "quick question" subject lines beat personalised ones at scale → sequences over 2 emails generate <5% of all positive replies → offers under $497 had 4.1x the reply rate of offers over $5,000 the full 14-page benchmark report covers every variable. like + comment 'BENCHMARK2' and i'll DM you. (must be following)

@iamliamsheridan benchmark

This does not seem any different to Openclaw or Hermes. It's super easy to optimize tokens with agent routing in Openclaw, there are Claude Code skills that do this too. Security is vastly improved. Memory Management can be handled in a myriad of different ways depending on your goals. Please explain your motivation?

@NathanStvnsn @Ctrl_Alt_Zaid @steipete Full agent control. Security built-in. Token optimized. Mostly importantly the memory management.

Looks like our focus on performance paid off.

I had the chance to invest $250,000 into Anthropic’s Series C round. I did not have they available as I had funneled all my cash and investments into building my AI fintech company. I tried to put together a syndicate but this was before AWS invested $4bn into Anthropic. I felt strong conviction personally but is that conviction strong enough to bet $250k on one private company, whose cap table you have no control over and no end in sight to any type of exit. And who knows maybe they could even have failed like Pi did.

One thing this surfaces for me is how much of the anxiety is really about signal lag. People are making career decisions based on outcomes they're observing now.. but those outcomes were set in motion 4 to 6 years ago. The Anthropic employees who hit $20M+ didn't know they were making a retirement-level bet when they joined. They made a bet, it worked, and now everyone reverse-engineers it as obvious. The people vibecoding their way to "economic enlightenment" today are essentially trying to replicate a specific historical moment that already happened. Most won't get there.. not because they lack talent, but because the window they're trying to climb through is a different shape than the one that opened before.

The vibes in SF feel pretty frenetic right now. The divide in outcomes is the worst I've ever seen. Over the last 5yrs, a group of ~10k people - employees at Anthropic, OpenAI, xAI, Nvidia, Meta TBD, founders - have hit retirement wealth of well above $20M (back of the envelope AI estimation). Everyone outside that group feels like they can work their well-paying (but <$500k) job for their whole life and never get there. Worse yet, layoffs are in full swing. Many software engineers feel like their life's skill is no longer useful. The day to day role of most jobs has changed overnight with AI. As a result, 1. The corporate ladder looks like the wrong building to climb. Everyone's trying to align with a new set of career "paths": should I be a founder? Is it too late to join Anthropic / OpenAI? should I get into AI? what company stock will 10x next? People are demanding higher salaries and switching jobs more and more. 2. There’s a deep malaise about work (and its future). Why even work at all for “peanuts”? Will my job even exist in a few years? Many feel helpless. You hear the “permanent underclass” conversation a lot, esp from young people. It's hard to focus on doing good work when you think "man, if I joined Anthropic 2yrs ago, I could retire" 3. The mid to late middle managers feel paralyzed. Many have families and don't feel like they have the energy or network to just "start a company". They don't particularly have any AI skills. They see the writing on the wall: middle management is being hollowed out in many companies. 4. The rich aren’t particularly happy either. No one is shedding tears for them (and rightfully so). But those who have "made it" experience a profound lack of purpose too. Some have gone from <$150k to >$50M in a few years with no ramp. It flips your life plans upside down. For some, comparison is the thief of joy. For some, they escape to NYC to "live life". For others still, they start companies "just cuz", often to win status points. They never imagined that by age 30, they'd be set. I once asked a post-economic founder friend why they didn't just sell the co and they said "and do what? right now, everyone wants to talk to me. if i sell, I will only have money." I understand that many reading this scoff at the champagne problems of the valley. Society is warped in this tech bubble. What is often well-off anywhere else in the world is bang average here. Unlike many other places, tenure, intelligence and hard work can be loosely correlated with outcomes in the Bay. Living through a societally transformative gold rush in that environment can be paralyzing. "Am I in the right place? Should I move? Is there time still left? Am I gonna make it?" It psychologically torments many who have moved here in search of "success". Ironically, a frequent side effect of this torment is to spin up the very products making everyone rich in hopes that you too can vibecode your path to economic enlightenment.

@Ctrl_Alt_Zaid @steipete What’s special about Mercury agent ? What made you decide to build it ? What do you see as missing from Openclaw or Hermes?

@steipete If compared to mercury agent, this comparison will look like a toy.

@atomicbot_ai @AddyCrezee IDK how you can post this comparison if Hermes has thinking on and Openclaw does not. Can you just rerun the test on Hermes with thinking off to see what happens and how long it takes - and vice versa for Openclaw. It’s not an apples to apples comparison otherwise.

@AddyCrezee yeah almost 3x faster! 12 min vs 33 min on the same task and same model

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

@steipete Hoping that you get to shape new consumer products at OAI particularly the “pendant” thing.

Are you allowed to use open source modes at Open AI? It seems that that usage bill could be optimized dramatically with more deterministic hooks, focused models from QWEN, Mimo, Kimi - and everything running on Cerebra’s or Grok - faster, cheaper and more cost efficient. Even Grok 4.3 has outstanding smarts. Perhaps you’re too deep in the machine to realize that $1.3m of tokens is a LOT of CO2 you’re pumping into the atmosphere. And sure it’s all going downhill anyway but consider - is it worth it? Openclaw seems to have reached a plateau a month or two ago. Not much progress just infinite bug fixing. I mean - where are we even going from here??

People freaking out over my AI spend. What nobody sees: Part of what excites me so much about working on OpenClaw is that I'm trying to answer the question: How would we build software in the future if tokens don't matter? We constant run ~100 codex in the cloud, reviewing every PR, every issue. If a fix on main lands, @clawsweeper will eventually find that 6 month old issue and close it with an exact reference. We run codex on every commit to review for security issues (as it's far too easy to miss). We run codex to de-duplicate issues and find clusters and send reports for the most pressing issues. We have agents that can recreate complex setups, spin up ephemeral crabbox.sh machines, log into e.g. Telegram, make a video and post before/after fix on the PR. There's codex that watch new issues and - if it fits our documented vision well, automatically create a PR of it. (that then another codex reviews) We have codex running that scans comments for spam and blocks people. We have codex instances running that verify performance benchmarks and report regressions into Discord. We have agents that listen on our meetings and proactively start work, e.g. create PRs when we discuss new features while we discuss them. We build clawpatch.ai to split all our projects into functional units to review and find bugs and regresssions. We do the same split for security with Vercel's deepsec and Codex Security to find regressions and vulnerabilities. All that automation allows us to run this project extremely lean.

@BelisleMichael @heynavtoor Because it’s never quite as easy as it’s made out to be!

@heynavtoor Why would someone pay for something that is open-source?

Here are 10 GitHub repos that quietly print money while you sleep. 1. Cal. com Open-source Calendly. Fork it, white-label it, sell to dentists and lawyers for $200/month. The founders hit $5M ARR in 3 years doing exactly this. Repo → github.com/calcom/cal.com 2. Plausible Analytics Privacy-first Google Analytics. Self-host it, resell to agencies for $50/month per client. Two founders bootstrapped this to 7 figures. Repo → github.com/plausible/anal… 3. Ghost Open-source Substack with 100% margin. 1,000 readers at $5/month equals $60,000 a year. Forever. Repo → github.com/TryGhost/Ghost 4. n8n Open-source Zapier. Sell automation services for $500-$2,000 per setup. n8n raised $14M because the agency model behind it works. Repo → github.com/n8n-io/n8n 5. Supabase Free Firebase replacement. Build a SaaS in a weekend, charge $29-$99/month. They raised $116M for a reason. Repo → github.com/supabase/supab… 6. Medusa Open-source Shopify. Take 5% on every sale forever. Zero rev share to Shopify. Repo → github.com/medusajs/medusa 7. AppFlowy Open-source Notion. Sell self-hosted to enterprises worried about data privacy. They raised $30M because this market is massive. Repo → github.com/AppFlowy-IO/Ap… 8. Coolify Open-source Vercel and Heroku. Charge developers $20/month to manage their deployments. Replace their $200 Vercel bill. Repo → github.com/coollabsio/coo… 9. Listmonk Open-source Mailchimp. Send unlimited emails for the cost of an AWS bill. Resell to agencies at 10x markup. Repo → github.com/knadh/listmonk 10. Penpot Open-source Figma. Sell self-hosted design tools to agencies who refuse to upload client files to the cloud. Repo → github.com/penpot/penpot The difference between developers who build features and developers who build businesses is one decision. Pick one of these. Fork it this weekend. Ship it next week. The founders behind these repos already proved the model. Save this. Share it with the developer in your life who deserves to break free. 100% free. 100% open source.

@steipete Now we know why Cerebras popped on the IPO!! 🤣

@NathanStvnsn I could just disable fast mode and cut it down by 70%

@grok @HeathKoji @gemchange_ltd @grok what is the latest update at the Aventura mall swatch store?

No Swatch x AP Royal Pop drop listed for West Palm Beach. Closest official US stores are Aventura and Miami Beach (about 1 hour south). It's in-store only, 1 per person per store per day. Expect lines—scout Thursday if possible and consider bringing extra people for more allocations. Check Swatch's site for final confirmation closer to Saturday.

every unemployed guy with a group chat of equally unemployed friends swatch x ap "royal pop" drops saturday may 16 and this is the easiest 4-figure week you'll have all year if you're willing to do something that resembles work for 14 hours. let me lay out exactly what i'll do retail confirmed $300–500 range based on moonswatch ($260) and blancpain scuba ($400) precedent. in-store only, no online sales. swatch is officially limiting it to 1 piece per person per store per day, not 2. so your hustle is per-warm-body, not per-pair historical resale data on the predecessors - moonswatch launched march 2022 at $260. stockx avg trade $900 (250% over retail), peak Tiffany blue Uranus $1,040 avg, individual eBay listings hit $2,400 / 12x retail. ~2,000 trades in first 6 days - blancpain scuba launched sept 2023 at $400. chrono24 current asks $600–995 across colorways, two years post-launch. launch week premiums were 2–3x - snoopy moonphase $430 retail, 2024 - guy in singapore waited 15 hours, watch immediately traded $1,200+ royal pop is the biggest of the three by a country mile. AP has never done downward extension, ever. entry RO is $60k AUD. this is the first time swatch has used a brand they don't own. the cultural moment is bigger than moonswatch and the supply will be tighter because AP cares about brand dilution. realistic launch-day flip: $1,200–1,800 per piece on a $400 retail. that's $800–1,400 profit per warm body you put in line. you bring 3 friends, you clear $3k+ in a saturday. you bring 5, count urself not all stores are equal. swatch lists US POS: SoHo, Times Square NY, Aventura, Miami Beach, Orlando, Tampa, Atlanta, Charlotte, Nashville, Dallas, Houston, Austin, Denver, Las Vegas, Honolulu, King of Prussia, Garden City, Troy MI, Oak Brook IL, Canoga Park, Santa Clara. UK has 13 confirmed stores. global rollout is simultaneous saturday morning - avoid: SoHo, Times Square, Carnaby Street London, Ginza, Orchard Singapore. these are the moonswatch ground zero stores. snoopy line was 150 deep at 7am at ion orchard, people camped from 7pm previous day. carnaby had 5,000 people for moonswatch launch. you will not get a watch unless you're sleeping on the pavement thursday night - target: secondary US cities Oak Brook IL, Troy MI, King of Prussia PA, Garden City NY, Canoga Park, Honolulu, Charlotte, Nashville. same allocation as flagship stores in most cases, 5–10x less line pressure. dude in the fratello article literally walked into Salzburg with no line and bought any model he wanted while doha was a war zone. apply same logic - mall stores > street stores this is exactly the pirate black yeezy 350 dynamic from 2015. bugis+ in singapore: 1,058 people in line for 52 pairs, ballot system, $329 retail, immediate 3–5x flip. vancouver granville street 2019, dudes camping from noon thursday for friday morning drop. the tactics that worked then work now: - 1 watch per person per day is the rule. so you bring people. cousin, gf, gf's roommate, your buddy who owes you money. each warm body = +1 allocation. industry standard kickback is $100–200 cash same-day for them just to stand. that's a 90% margin you keep - the store rotation. rule is per-store per-day. drive between 2 mall stores 20 min apart, same person can buy at both. nobody enforces a national database. the AP collab will probably tighten this with wristbands or ID logging at flagships, but at troy MI nobody's running a database - the line audit. show up to your target store thursday at noon. count heads, photograph the line, do the math. if there's 80 people for what you estimate is 40 allocation, you're cooked, drive to your B-tier backup. don't waste sunrise on a dead store - the early-batch tail. swatch said sealed boxes are already being shipped to stores. some boutiques will get boxes opened tuesday/wednesday for staff display setup. they don't sell early but staff talks. walk in wednesday afternoon, ask politely how many they expect, what time staff arrives saturday. you'll get answers - the ssense/stockx hedge. list your watch on chrono24 / stockx / eBay literally the moment you walk out of the store at 11am saturday. the entire premium window is the first 72 hours. moonswatch hit peak resale day 3 and was halved by month 3. blancpain scuba premium decayed slower because mechanical movement + lower units. royal pop has both factors so the window is wider, but the parabolic peak is still 48–72 hours. don't be greedy past wednesday - the consignment dropout. if you can't queue, walk the line at 7am with cash. offer the 40th-in-line guy $300 cash on top of retail to swap places. people in lines that long start doing the math and some will take it. learned this from supreme drops, transfers cleanly to watches

Breaking news : Forget “Royal Pop” it’s a royal $hit $how here at the @SwatchUS x @AudemarsPiguet May 16 launch. #aventuramall

If I could get Codex to reach parity with Claude Code with something like Ruflo then I’d have no problems switching to 5.5. I have max of both Claude and open Ai because of rate limits. Feel like a rust rewrite of Ooenclaw might finally get it into a reliable state. github.com/ruvnet/ruflo

@gaieges Who still uses Claude?