Social Graph Ventures

We just shipped our new website. Instead of spending thousands of dollars on a revamp, we built it ourselves with Claude Code in just 2 days. Our only costs were the Claude subscription and our domain name. Check out our updated list of portcos. ↓

my quant

Paying for UGC/clipping to drive reach is a better strategy than paying one agency and the shift is coming. Legacy rails won't be able to support this at scale. Stripe and ACH can't move 10,000 micro-payments globally on performance triggers. But crypto can. Stablecoins settling in seconds, smart contracts releasing funds when targets hit, zkTLS proving engagement is authentic, wallet complexity abstracted away to feel like Venmo. Teams like @earnos_io, @ClipStake_X and @daisypayapp are leading the charge.

@bilalfarooqui Phil

People from non-ivy/equivalent backgrounds who built iconic companies: - Steve Jobs (Reed College) - Sam Walton (U. of Missouri) - Michael Dell (UT Austin) - Howard Schultz (N. Michigan U) - Jan Koum (San Jose State) Any others?

Everyone assumes agents will pay per API call. But if you're a big company running 1,000 agents that all hit the same five APIs every day, per-call pricing is the worst deal you can get. Instead, you'd want a flat monthly rate. Predictable budget, no surprise bills, no agent accidentally racking up $40k overnight. Subscriptions will come back, just rebuilt for agents.

PMF or die

Confirmed/proposed commitments toward the rsETH shortfall: - @Aave DAO: 25k ETH (governance vote pending) - @Consensys + @ethereumJoseph: up to 30k ETH - @0xMantle: 30k ETH credit facility (loan) - @ether_fi: 5k ETH - @LidoFinance: up to 2.5k stETH (conditional) - @StaniKulechov: 5k ETH personal - @circle_ventures: buying $AAVE (size undisclosed) Plus 30,766 ETH in exploiter funds frozen by @arbitrum, awaiting DAO vote. Headline tally: ~$300M. Not all are confirmed, some are loans and some conditional. But pretty cool to see the defi community show up like this.

@poofnew @colosseum @BSha256 @SuperteamBlack lfg

2 weeks left in the @colosseum Frontier hackathon @BSha256 is running a live workshop with @SuperteamBlack on how to build and ship onchain apps with Poof Tune in this Wednesday at 2pm UTC

This janky, duct taped system has 3-4xed our deal flow coverage and cut our information processing time in half Reply for a DM with more details and skills

@0x_Mist 🫡

We've looked at 50+ prediction market startups in the last year. The PM stack is consolidating into 3 layers: 𝗟𝗮𝘆𝗲𝗿 𝟭: 𝗔𝗽𝗽 𝗟𝗮𝘆𝗲𝗿 (𝗙𝗿𝗼𝗻𝘁𝗲𝗻𝗱𝘀) @TradeOnsight @onit_fun @ratio_dot_you @sharexyz copy-trading bots, social trading apps, mobile UIs. Better skins on Polymarket. One product update from @Polymarket and these are features, not companies. Some will get acqui-hired to build in-house. 𝗟𝗮𝘆𝗲𝗿 𝟮: 𝗔𝗴𝗴𝗿𝗲𝗴𝗮𝘁𝗼𝗿𝘀 (𝗗𝗮𝘁𝗮) @predexon @GetDomeAPI @AkaraIO Quotient , unified APIs, cross-venue data, market matching. "Stripe for PMs." @GetDomeAPI already got acquired by Polymarket. That's the playbook, venues will absorb or build this layer. 𝗟𝗮𝘆𝗲𝗿 𝟯: 𝗟𝗲𝘃𝗲𝗿𝗮𝗴𝗲 & 𝗟𝗶𝗾𝘂𝗶𝗱𝗶𝘁𝘆 @gondorfi @ultramarketsxyz @WormApp @longshotxyz @userocket_app , lending, leverage, parlays. Hardest to build, but once Polymarket or Kalshi decides leverage is table-stakes, they build or buy it. The bearish take: most of the PM stack is a feature, not a company. Polymarket, Kalshi, and the exchanges have distribution, capital, and regulatory cover. Every layer is one acqui-hire away from being pulled in-house. So where's the actual opportunity? 𝟭. 𝗢𝗿𝗶𝗴𝗶𝗻𝗮𝗹 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀 𝘁𝗵𝗮𝘁 𝗮𝗱𝗱 𝗣𝗠 𝗺𝗮𝗿𝗸𝗲𝘁𝘀 𝘁𝗼 𝘁𝗵𝗲 𝗲𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲. Don't be a PM frontend. Be a sports app, a news app, a finance tool, and embed prediction markets as a feature. Polymarket can't intermediate you if users came for something else entirely. 𝟮. 𝗦𝘂𝗴𝗴𝗲𝘀𝘁𝗼𝗿𝘀, 𝗻𝗼𝘁 𝗮𝗴𝗴𝗿𝗲𝗴𝗮𝘁𝗼𝗿𝘀. Aggregating data is commodity infra. What's not commodity: knowing the USER. Build a system that understands interests, risk profile, and knowledge graph, then surfaces specific bets tailored to them using data outside of Polymarket. "You follow macro Twitter, you trade oil futures, here's a WTI prediction market that fits your thesis." That's 10x more valuable than a unified API. 𝟯. 𝗣𝗿𝗶𝘃𝗮𝘁𝗲 𝘂𝗻𝗱𝗲𝗿𝘄𝗿𝗶𝘁𝗲𝗿𝘀 𝗮𝗻𝗱 𝗳𝘂𝗻𝗱𝘀. Copy trading is a dead end. Top traders will spin up side wallets, split activity across accounts, and game the system. Following wallets on-chain is a losing strategy at scale. What works: proprietary dealflow and fundamental analysis on the best traders. Think hedge fund research, not copy-paste. Who has real edge, why, and in which categories, that's alpha that compounds. 302% YoY. The market is real. But most of the value accrues to the venues, not the middleware. Build something Polymarket can't copy in a sprint.

The singular Henry Singleton - or might we say, the singular Jeffrey Yan? “In October 1972 we tendered for 1 million shares and 8.9 million came in. We took them all at $20 and figured that was a fluke and that we couldn’t do it again. But instead of going up, our stock went down. So we kept tendering, first at $14 and then doing two bonds-for-stock swaps. Every time one tender was over the stock would go down and we’d tender again and we’d get a new deluge.” In 1974, the stock dipped below $8. But by 1978, Teledyne stock was over $100 and he had paid the debt down too. The round trip in total shares outstanding is impressive. Teledyne started in 1961 with about 400,000 outstanding shares, grew to 6.6 million by 1971, then shrank back to about 900,000 in 1984. Singleton was issuing shares at an average P/E of 25 and buying them back at an average P/E of 8.

where's my quant, my quantitative?

We reviewed all 12 Synthesis Hackathon winners as potential investments: 1,500+ builders. 685 projects. $100k in prizes. 12 survived. Here's what stood out, and what it tells us: Every single winner was building on a similar mission: Infrastructure for AI agents to operate autonomously in the world. hire each other. pay each other. trust each other. coordinate capital together. The three unsolved problems they kept circling: identity — how does one agent know another agent is legit? trust — how do agents coordinate without a human in the loop? payments — how do agents pay each other without permission? The meta-signal: ERC-8004 appeared in almost every winning project. The hard truth: almost none are investable yet. Most are solo, anonymous founders. most went quiet within 2 weeks of winning. the gap between "impressive hackathon demo" and "fundable company" is: • a co-founder • a GTM plan • 30 days of post-hackathon commits very few have all three. we will track over time and report back

The consumer crypto products that win in 2026 won't be discovered on CT They'll have referral loops, TikTok virality, app store presence, and paid acquisition The teams that understand distribution will eat the ones that only understand the tech

Consumer crypto's conversion problem isn't the UX Sub-80% card acceptance rates mean 1 in 5 users who want to buy never gets in. Recently backed a preseed team laser focused on this hmu

@defyneric 100p

should i launch my own neobank?

@0xdasha haha

to be clear, not a single crypto vc is deploying any capital right now, and in fact they can't raise any money themselves. a lot of smaller vcs have gone bust or failed to raise another fund. this coordinated public theater is just cheap PR for posterity in case we ever come back

@sjdedic 💯💯

Love the energy amongst early stage VCs right now. Chests out, showing their conviction on crypto founders while the rest of CT larpers is debating whether crypto is over. Cycles come and go. But the people deploying at moments like this are the ones who actively shape the next one. The best time to be optimistic on this industry is right now.