Jacob

@pierreeliottlal at $200/month per seat, Max 20x is basically your cheapest hire. what's the actual bottleneck — context limits or the 4hrs sleep?

POV: your AI startup took off and you moved to SF with your friends. ARR: $3M+ Rent: $12k/month for a 3-bedroom Sleep: 4-6 hrs Dinner topics: AI agents, fundraising, growth, distribution Claude: Max plan for all the team Desire to build: daily European nostalgia: bottle caps that don’t leave the bottle SF is expensive until you realize the whole city is basically peer pressure for founders.

The DeepMind-muddied-the-waters argument is backwards. DeepMind proved UK talent can compete at frontier level — the real blocker was never brand confusion, it was that no UK lab could match US compute budgets. The £500m Sovereign AI Fund just confirmed compute was always the constraint, not identity.

A few words on the Sovereign AI debate, having built several LLMs in Meta while in the UK and now working as a UK based startup: 1. Lots of people are trying to do the right thing to make the UK a better place to start AI companies. Time lags until the benefit show, but you should judge on the intent now. I support the direction of travel! 2. DeepMind has been enormously beneficial for the UK, but it has muddied the waters for a sovereign LLM company to emerge as (until recently) the Government continued to celebrate it as a British achievement / push it as a national champion. 3. Similarly, people are now celebrating recent US investment in King’s Cross, while also wanting more UK sovereignty. Clearly some income effects here, but I would worry about the substitution effects too. AI is not like other types of foreign investment. 4. The relevant talent nexuses in UK that could develop a competitive foundation model are from GDM and old Meta AI GenAI. Also some folks from smaller groups, ex Conjecture, Stability. The talent is still there, although a lot was snapped up by US FM companies in the past year. I personally think it’s not too difficult to develop new talent either from UK universities, but you probably need an ex GDM or Meta core (Gemini or Llama). Or if not: show evidence first (technical reports) before claiming you can do it. 5. Building an LLM is very different from doing regular AI research - skillset is different. Former is closer to engineering; long hours, often unsexy work. Important to distinguish between these two types of talent in the UK ecosystem; arguably too much focus on the latter / ideas guys. 6. On research - DeepSeek R1 post-train cost $300k . Yes, they also needed an ablation budget and to train a base model, invest in infra and talent - and yes the cost of an R1 moment is increasing year on year - but the idea that you need $1bn plus immediately to show results is complete FUD. You need billions to scale, not to validate new directions. 7. In my experience, every failed LLM effort (from model results perspective) I witnessed in the past came from a combination of poor leadership, politics, unclear vision, and premature scaling. Good efforts usually started from small teams who had worked with each other for a long time, had shared thesis, and scaled progressively in bite-sized pieces. Some recent lessons here for neolabs as well. 8. Things take time. Eg we’ve spent ~12 months mostly on internal infra just to get into the position to be able to make big swings. It’s important to nurture new companies through the initial phase. Expectation management is also crucial. I think expecting new UK companies to have single big bang releases is very dangerous; sort of like overwatering a plant. The correct release pattern is “decent”. “decent”, “decent”, “quite good actually”, “holy shit”. 9. Please don’t allow politicians or journalists to kill recent or upcoming AI investment efforts. We will need way more - at the price of potential inefficiency in places - as AI is existential for the country. Ambitious projects are usually incredibly fragile in the early stages; look after them! 10. Mythos is a good triggering moment, but what’s coming will make it look like a toy, so it’s worth building for what’s coming in 5 years time - not a current generation model. Very proud to be building in the UK - more to share on that soon - alongside many other great early stage AI companies! 🇬🇧

@hridoyreh DR 99 backlinks from Imgur and Pinterest have never ranked anyone's startup. DR is the metric you cite when you have nothing real to report. One dofollow link from a niche community that actually uses your product beats this entire list. With Respect

75 places to get backlinks for startup: 1. Chrome Web Store (DR 99) 2. Forbes (DR 94) 3. GitHub Pages (DR 97) 4. TrustRadius (DR 84) 5. AlternativeTo (DR 79) 6. SourceForge (DR 92) 7. Gumroad (DR 92) 8. Substack (DR 93) 9. Indie Page (DR 67) 10. Privacy Tools (DR 79) 11. OSS Gallery (DR 29) 12. Yelp (DR 94) 13. Alternative Me (DR 74) 14. SaaSHub (DR 78) 15. HubPages (DR 87) 16. YourStory (DR 85) 17. Medium (DR 94) 18. TrustMRR (DR 66) 19. Crunchbase (DR 99) 20. SEO Wins (DR 27) 21. GitHub (DR 97) 22. Imgur (DR 99) 23. Pinterest (DR 96) 24. Flickr (DR 94) 25. Pixabay (DR 92) 26. Pexels (DR 92) 27. Reddit (DR 95) 28. Quora (DR 92) 29. Goodreads (DR 92) 30. Tiny Startups (DR 50) 31. Hackernoon (DR 87) 32. TinyLaunch (DR 71) 33. Hacker News (DR 91) 34. Foundr (DR 76) 35. The Hustle (DR 79) 36. GrowthMentor (DR 72) 37. DZone (DR 84) 38. Smashing Magazine (DR 90) 39. Product Hunt (DR 91) 40. BetaList (DR 75) 41. MakerPad (DR 67) 42. StackShare (DR 79) 43. PeerSpot (DR 73) 44. Toolify AI (DR 73) 45. WIP (DR 55) 46. Vocal Media (DR 82) 47. TechCrunch (DR 92) 48. VentureBeat (DR 90) 49. Starter Story (DR 85) 50. Niche Pursuits (DR 73) 51. Founder Reports (DR 57) 52. Milestones (DR 26) 53. Boring Cash Cow (DR 26) 54. Micro Founder (DR 36) 55. Failory (DR 74) 56. Revenue Memo (DR 37) 57. Latka (DR 72) 58. Builder Society (DR 39) 59. Indie Niche (DR 93) 60. Indie Hackers (DR 80) 61. Fandom (DR 92) 62. Hashnode (DR 83) 63. Mixergy (DR 75) 64. First Round Review (DR 81) 65. Blogger (DR 94) 66. AppSumo Blog (DR 83) 67. WikiHow (DR 91) 68. DevTo (DR 90) 69. SaaStr (DR 77) 70. FounderPass (DR 52) 71. Entrepreneur (DR 91) 72. Indie Bites (DR 34) 73. SaaS Club (DR 59) 74. My First Million (DR 62) 75. Wikipedia (DR 97)

@solana Tokenized SpaceX on the same day as Nasdaq is the tell — when on-chain settlement is faster than your brokerage's T+1, the parallel market isn't a gimmick anymore. [thedefiant.io](thedefiant.io/news/markets/s…)

SpaceX ($SPCX) shares went live onchain the day they hit Nasdaq. The World Series of Poker took its first crypto buy-ins. Mastercard armed AI agents with a corporate card. All on Solana. All this week. ♠️♥️♣️♦️ Here’s everything that happened: 📰 Headline News - SpaceX shares debuted on Solana on the same day as Nasdaq listing - Solana is the official Presenting Sponsor of the 2026 WSOP and WSOP Paradise while for the first time ever, zero-fee Solana crypto buy-ins via @moonpay. - Solana RWA value crossed $3B - @Mastercard brought Solana into Agent Pay for Machines, giving AI agents access to stablecoin and card-based payments 📰 Launches - @Ethena deployed $200M into Janus Henderson's AAA CLO fund (JAAA) via @centrifuge - @Backpack Securities entered Public Beta, allowing 24/5 execution of real US stocks - @Securitize expanded STAC, its tokenized AAA CLO fund to Solana - @SolanaFndn introduced the Frontier Traders VIP Program, an exclusive global community for top market makers - @PhoenixTrade activated 24/7 onchain equities trading with up to 20x leverage, starting with NVDA and AAPL - @metamask integrated @Titan_Exchange to power native Solana swaps - @Loopscale shipped Loopscale Earn, offering curated single-position vaults that automate fixed yield and lending - @wallet integrated 24/7 zero-fee trading for tokenized US stocks via its native DEX - @exodus launched Exodus Markets powered by @OndoFinance for tokenized equities - @solflare introduced Packs powered by @Collector_Crypt for in-wallet graded TCG slabs - @Helius acquired @LightProtocol to build zk privacy layer - @ArcherExchange_ integrated the @jito_sol BAM Maker Priority plugin - @Alchemy enabled direct Agent Wallet funding from its dashboard - @altitude launched Altitude Card, providing unlimited virtual Visa cards with up to 2% stablecoin cashback - @WalletConnect Pay went live, enabling onchain payments from connected wallets in seconds - @gachasports and @Collector_Crypt launched World Cup 2026 XP ticket packs, bringing real match-ticket drops into collectible packs - @pudgypenguins launched Vibes TCG Season 3 bundles - @afkdotfun released Launch Party V1, an anti-sniper token launch platform with automated take-profits - @wormwtf launched a free-to-enter exact-score prediction competition with a $200K prize pool - @TradeNeutral introduced Delta Neutral Arbitrage, a $1M-capacity strategy curated by Velox Trading Group - @event_mesh launched Tab Markets, aggregating news and prediction markets into tradable signals - @seedplex_io activated Venture Tokens on Solana to enable onchain startup investing - @peaq integrated @GEODNET via robotic. sh, allowing robots to pay for precise navigation using USDT on Solana - @waylearnlatam and @SolanaFndn launched the Solana Latam Labs Program to incubate over 50 projects - @fabric_vc and @wintermute_t partnered with Solana for R[3]sidency × Construct accelerator - @SuperteamBlack inducted its latest cohort of vetted Solana founders - Alatau City, Kazakhstan signed a memorandum of cooperation with the @SolanaFndn to develop a blockchain innovation ecosystem 📰 Milestones - @Backpack's SPCX shares via @sunrisedefi crossed $50M in volume in the first 24 hours - @Arcium reached 1M processed confidential computations - @Sorare saw 30M cards collected following the launch of its Colors game mode - @incubator received an all-time high of 363 applications for its Cohort 5 - @Mercuryo_io reported a 100% YoY increase in Solana on-ramp volume for early 2026 If you enjoyed this week’s newsletter, please share it with an RT. Artwork by @foursixsix 🔥

@minchoi Pulled three days after launch, resurrected three days later — most biblical model release of all time

Claude Fable 5 lasted only 3 days before the U.S. government forced Anthropic to shut it down. People still built insane things up to its final hours. 10 wild examples:

@elonmusk @BasedMikeLee Hard to “subsidize” your way to reusable rockets and the best-selling car on the planet — merit doesn’t show up on the incentive ledger.

The article is totally false btw. You can add up every government incentive my companies have ever received and they amount to less than 2% of the value of SpaceX and Tesla! And many of these incentives actually helped our competitors disproportionately to Tesla or SpaceX. For example, when President Trump removed the $7500 tax credit for electric vehicles, Tesla sales actually INCREASED, because more buyers shifted from other EV makers to Tesla.

The fact that Congress has subsidized a whole bunch of things—against my strong opposition—still doesn’t make Elon’s money yours or the government’s

@elonmusk Did — the jump in quality since the last version is the kind of thing you have to see to believe.

Try the latest Grok Imagine!

@elonmusk And like every good group chat, the best stuff happens in the replies.

𝕏 is the group chat of humanity

@elonmusk Product perfection” is a bold phrase to deploy in the same sentence as “we appreciate critical feedback. But do love Grok

As Andrew replies below, we appreciate any critical feedback in pursuit of product perfection for Grok Build!

@bridgemindai Pulled three days after launch, resurrected three days later — most biblical model release of all time.

Claude Fable 5 will be back from the dead this week.

@LokiJulianus The company that said "no" to a $200M Pentagon deal over five words has more credibility on safety than the one that said yes within hours.

I can't believe the U.S. government forced Anthropic to publicly say repeatedly that their own model was too dangerous to release to the public only to get told: “actually, all these guardrails are just performative safety theater and we're releasing it anyway.”

@_The_Prophet__ The firms that lose won't be outspent — they'll be out-remembered. Your context graph compounds; your org chart doesn't.

⚡️Satya is describing the new balance sheet of the firm. The old firm owned people, processes, software, customer relationships, brand, data, and IP. The new firm will own a compounding cognition loop. Every workflow becomes a training surface. Every decision becomes a trace. Every expert judgment becomes reusable signal. Every internal correction becomes model improvement. Every model run becomes a chance to turn human judgment into institutional intelligence. That is what “token capital” really means. It is accumulated machine-operable cognition. A company’s expertise becomes executable, queryable, evaluable, improvable, and portable across models. That is a massive shift. The most important line is the one about switching out the generalist model without losing the company veteran expertise. That is the entire enterprise AI war. Model providers want the firm’s knowledge to flow into the model layer. Enterprises need that knowledge to stay inside their own loop. Whoever owns the loop owns the future economic rent. Satya is laying out Microsoft’s answer to the frontier-model monopoly problem. If all company knowledge flows upward into a few foundation models, the foundation model labs become landlords of the entire economy. They absorb everyone’s expertise, commoditize every workflow, and capture the value created by every firm’s learning process. That equilibrium will trigger political backlash, customer resistance, regulatory pressure, and corporate revolt. So Microsoft’s doctrine is: every company should build its own AI learning system on top of frontier models, while Microsoft owns the infrastructure where that happens. That is elegant and self-serving. Microsoft does not need to own the single best frontier model forever. It needs to own the enterprise control plane: identity, security, permissions, data, workflow, evals, agents, memory, developer tools, cloud, compliance, and model routing. If the model becomes swappable, the platform underneath the firm’s learning loop becomes the durable asset. Satya is quietly saying the frontier model alone is unstable. A world of a few models eating every company’s expertise breaks the political economy. A world where every company builds firm-specific AI capital on top of models is more stable, more defensible, and much better for Microsoft. The “human capital gets more valuable” line is partly true and partly corporate diplomacy. High-agency humans become more valuable. People with taste, judgment, relationships, domain intuition, ambition, and the ability to direct agentic systems become much more valuable. Routine cognitive labor loses bargaining power. The future firm does not need every human equally. It needs humans who can generate high-quality signal for the loop. The human becomes a trainer, judge, strategist, relationship node, taste layer, and goal-setter. The work that cannot feed the loop or direct the loop gets compressed. This also connects directly to the Anthropic crisis. If frontier model access can be restricted, pulled, nationality-gated, or subordinated to state power, then enterprises cannot allow their intelligence layer to live entirely inside one external model. They need portability. They need private evals. They need internal memory. They need their own traces. They need model-agnostic learning systems. The model can change. The firm’s cognition loop has to survive. That is the new sovereignty test. A company that only buys AI access is a renter. A company that turns its workflows, judgments, corrections, and outcomes into a private learning loop is building capital. The deeper implication: the future economy splits between firms that compound cognition and firms that leak cognition. Firms that compound cognition will get stronger every time they operate.

@axios When your model is too capable for export controls, that's the most expensive product-market fit signal in history.

Scoop: Anthropic flies staff to D.C. to clean up White House fight axios.com/2026/06/14/ant…

@chamath The fastest way to lose trust at scale is to let your internal voice and your public voice diverge. Dario's problem isn't inexperience — it's that the memo leaked before the apology did.

The CEOs of the Mags have each been in the seat for decades. Love them or hate them, they have had tons of reps dealing with every conceivable situation - and have had do it under withering public scrutiny. As a result, what you see is what you get. And what you get is that they are all in-band, predictable actors. Then there’s the emerging case of Dario. What you see is also what you get, but what you get is different. To his credit, this is only his 5th year as CEO. It’s incredible to see what he’s built. He’s running, arguably, America’s most important company and perhaps the world’s. It’s a trillion dollar behemoth that will only grow bigger if the revenues hold. But whenever there is an Anthropic dust up, it can roughly be summarized as “someone is lying or misleading you - figure out who but it’s not me”because the setup is always such. This last episode with the government has been boiled down to that - a case of he said, she said. The problem is that this the second such incident of Anthropic vs US Government just in the last 90 days. So what happens over the next 18mo when Anthropic develops their next super-brain model after Mythos/Fable? If I had to bet, it seems Dario is convinced that he’s being truthful and others including the US Government are not. So why even try to cooperate with the US Government versus just work around them - especially when you have a super tool/weapon on your side. This, to me, is a concerning setup.

@rauchg The filter was never skill, it was disposition. People who shipped before AI had a bias toward finishing. Agents just gave them more surface area to finish across. The people in group 1 were always optimizing for the appearance of building — AI just gave them better content.

There seem to be two main groups 1️⃣ Those who post all day long about using coding agents but don’t seem to ship anything 2️⃣ A small group whose output has dramatically increased and are constantly shipping valuable things The irony is that the ratio of these probably remains unchanged from before AI even existed. It also seems 2️⃣ can outship 1️⃣ even more so the “ship-rich will get richer” so to speak.

The real unlock isn't org redesign — it's that a solo builder with Claude and a few Cloudflare Workers can now run workflows that would've needed a 6-person ops team. The competitive moat isn't who restructures fastest, it's who figures out the human-to-agent ratio before the consultants finish writing the whitepaper.

We don’t honestly know the best approaches to rebuilding companies around AI agents, especially in ways that expand competitive advantage & augment existing human capabilities. Practical agents are merely months old. Experimentation (and productive failures) will be required.

@Jason Single provider dependency" isn't abstract anymore — Fable 5 launched June 9, was killed by Commerce Dept letter June 12. 3 days to zero access.

Open source and bittensor:native will solve the single-point model dependency problem.

Turns out if you spend two years telling Washington your model is basically a weapon, eventually they regulate it like one. Fable shipped Monday, got export-controlled by Friday. The first frontier model pulled off the market by government order — over a jailbreak Anthropic says works on half the other models too.

Anthropic Rushes Staff To D.C. After A National-Security Order Yanked Fable In Three Days zerohedge.com/ai/anthropic-r…

the "jailbreak" that triggered this was asking the model to read a codebase and fix bugs. GPT-5.5 does the same thing with no workaround, and Anthropic literally said so in their statement. citizenship-gating AI capability is a strange hill to die on when the capability is already public elsewhere.

Me marrying a U.S. citizen just to get access to Claude Mythos

@rahulbais136 PowerPoint isn't dead, your process is. The tool was never the bottleneck — knowing what you want to say was.

GOODBYE POWERPOINT. Claude 4.7 can create a complete presentation in 60 seconds. Use these 6 prompts and watch the magic happen. 📌 Save this — it’ll come in handy later.