Pinned Tweet
In 2024, a company I built and sold ended up on a Times Square billboard in New York City
In early 2023, I was experimenting a lot with ChatGPT. Then, in March 2023, I got access to OpenAI’s first ChatGPT API release.
I traveled a lot and always found trip planning unnecessarily painful. I wanted something that understood my preferences, where I wanted to go, what I liked to eat, and how I liked to travel, then turned that into a personalized itinerary.
So I built it.
At the time, you couldn’t simply prompt this into existence.
I spent weeks building the product, searching Stack Overflow when things broke, stitching together APIs, and figuring out how to turn an LLM into an actual product.
I launched it on Product Hunt.
Within a few months, it grew to more than 10,000 users.
Travel companies started reaching out about integrating the technology into their platforms.
In early 2024, I listed the company for sale on @acquiredotcom .
The business was profitable and high margin, but I knew I could not give it the time required to take it to the next stage.
A few weeks later, I sold it to an Entrepreneur and Investor based in USA.
Then came a pretty surreal moment.
The founder of @acquiredotcom featured the acquisition on a Times Square billboard in New York City.
Seeing something I had started a year earlier displayed in NYC Times Square was one of those founder moments I will not forget.
But what I find even more interesting is what happened next.
Fast forward roughly two years and the core experience I spent weeks building can now largely be reproduced with a single prompt.
AI can understand your preferences, research a destination, and create a personalized itinerary in seconds.
Rebuilding the product itself is even more dramatic.
In 2023, building meant writing code, searching Stack Overflow, debugging integrations, and spending weeks getting an MVP into people's hands.
Today, you can describe a product with your voice and have AI generate the application without writing a single line of code yourself.
That changes the economics of building software.
The cost of going from idea to product is collapsing.
Technical execution is becoming increasingly commoditized.
Which means two things matter more:
Distribution and moats.
Distribution was always hard. Now it is arguably the main problem to solve.
- Find the right users.
- Get in front of them.
- Build trust and brand.
- Give them reasons to come back.
But distribution alone is not enough.
If your product can be replicated with a prompt or rebuilt in a weekend, what makes it defensible?
Your moat might be proprietary data, network effects, distribution, deep workflow integration, switching costs, brand, community, or unique supply.
The bar for building software has never been lower.
But the bar for building a durable company is actually getting higher.
AI has made it dramatically easier to build.
It has not made it dramatically easier to matter.
Thanks again @agazdecki
Riz retweeted
Limit orders are live in MetaMask.
Set your price on a token swap and let us do the rest! ⤵️
Riz retweeted
Seoul, Tokyo, Singapore 🦊💕
Finally home. Caught up on sleep, and my voice is slowly coming back after so many good conversations.
Some of the sharpest traders, builders, and creators I know are in Asia. On this trip, many of them joined us for long dinners, gave us honest feedback, and helped shape what we’re building toward Open Money.
I learned so much from them, and I’d love to do this more often.
If you’re a serious trader, creator, or community builder in Japan or Korea and want to do more with MetaMask, whether that’s sharing product feedback, joining our next event, or creating content together, get in touch below! 🫶
Special thanks to our Gold Fox @Edward__Park and friends for helping MetaMask win and improve our products 🙌 @hyeon__dev @dorong_x @flarrs315 @ditto_anc @minebuu @JuhyukB
너무너무 고마운 사람들 @tylerrasch @100y_eth @aquiahoramism0 @magonia_B @Hope_web3_ @BQYouTube @HighballCharlie for all the support at our events, and my sweet fox fam @0xAYoo @woohyun_chong @Dora_web4 @0xtomsargent @francescoswiss @spencecoin 사랑해요~ 🦊
so let me get this straight…
people are opening 1000x long/short positions and losing it all on purpose to get $PAPER so they will be able to stake asap to share the $15m and growing pot of staking rewards as more people lose even more money to fund those staking rewards?
hello??? which part of the cycle is this?
Crypto social trading has a trust problem. And better leaderboards alone won’t solve it.
I’ve been following the evolution of social trading across platforms like FOMO, PumpFun, Axiom, Legend, and Invo.
The category is growing rapidly, but looking at user feedback reveals a recurring set of problems.
Most come down to three things: transparency, incentive alignment, and execution.
And I think solving them will define the next generation of social trading.
1. A profitable trader isn’t necessarily a profitable trader to follow.
This is probably the most important distinction in social trading.
A trader might generate exceptional returns, but that doesn’t mean their followers can replicate them.
Especially in memecoin markets, where returns are often power-law distributed.
A small number of trades can generate multiples large enough to offset dozens of losing positions.
That’s why conventional metrics like win rate can be misleading.
What matters is the actual distribution of realized returns, the magnitude of winners relative to losers, capital deployed, drawdowns, and the degree to which performance depends on a handful of outliers.
But even those metrics only tell half the story.
A trader who entered at a $50K market cap and exited at $5M might have generated an extraordinary return.
Someone copying that trade at $2M is participating in an entirely different risk/reward opportunity.
The leader’s PnL and the follower’s opportunity are not the same thing.
2. Copy trading introduces a second layer of risk: execution.
Memecoin trading is extremely sensitive to timing, liquidity, and position sizing.
A profitable strategy can become unprofitable when copied at worse prices.
And the more people following the same trader, the more challenging execution can become.
Consider a trader with a strong historical track record.
If thousands of followers attempt to enter the same low-liquidity token, their collective demand can materially move the price.
Some followers get favorable entries. Others become exit liquidity.
The same problem applies when everyone tries to exit.
Copy trading therefore needs more than a simple follow button.
It needs safeguards around maximum entry deviation, price impact, liquidity, position sizing, and exits.
Ideally, users should understand how followers historically performed relative to the trader they copied.
3. On-chain transparency doesn’t automatically mean meaningful transparency.
Blockchains make transactions observable.
But interpreting those transactions is a different challenge.
A trader can distribute holdings across multiple wallets, accumulate before publishing a thesis, or sell through wallets that aren’t visibly associated with their public profile.
Similarly, a token might appear to have broad holder distribution while substantial supply is controlled by a cluster of connected wallets.
This creates information asymmetry.
For social trading to mature, platforms need to make these signals accessible.
Not just realized PnL and trade history, but also wallet clustering, concentration risk, entry timing, and potentially coordinated trading activity.
These are probabilistic risk signals, not definitive proof of manipulation, but they can materially improve decision-making.
3
1
160
4. Incentives matter more than engagement.
Social trading platforms naturally want more activity.
More followers, more callouts, more transactions.
But rewarding activity doesn’t necessarily reward good outcomes.
Recent controversy around creator rewards on PumpFun illustrates the tension between rewarding influence and rewarding trading performance.
If creators benefit primarily from generating volume or attracting buyers, their incentives can diverge from those of their followers.
The more interesting question is whether platforms can design incentive systems around genuine value creation.
For example, evaluating a trader’s published opportunities against actual executable prices, accounting for transaction costs, and measuring the outcomes available to followers.
That is considerably harder than counting clicks, volume, or headline returns.
But it’s also much more meaningful.
5. Security and execution are foundational to trust.
Recent complaints across trading platforms highlight issues ranging from wallet security and session compromises to failed transactions, unexpected execution, and limited exit functionality.
For traders, these aren’t secondary UX problems.
They’re fundamental product risks.
A platform can offer excellent discovery and powerful social features, but none of that matters if users cannot reliably manage or exit positions.
Transparent execution, predictable fees, robust security, and user control over assets need to be built into the foundation
Where I think social trading goes next
The first generation of social trading made trading more discoverable and accessible.
The next generation needs to make it more intelligent.
That means moving beyond simplistic leaderboards and viral PnL screenshots toward:
Performance transparency: Realized returns, return distributions, capital efficiency, drawdowns, and concentration of profits.
Execution-aware discovery: Understanding whether an opportunity remains attractive at the price and liquidity available to followers.
On-chain risk intelligence: Wallet clustering, holder concentration, and potential manipulation signals.
Aligned incentives: Rewarding genuine trading value rather than attention and transaction volume alone.
User-controlled risk: Better entry constraints, position sizing, exit controls, and self-custody.
The opportunity is much bigger than building another trading app with a social feed.
Social trading can fundamentally change how people discover opportunities, evaluate risk, and participate in markets.
But there’s a critical difference between discovering a trader who made money and discovering an opportunity where you can still make money.
I think the platforms that understand that distinction will ultimately win.
29
Riz retweeted
1/ Meet @0xEthan
He runs an extraction scheme on pumpfun
buy → callout → sell from side wallets
49 tokens. $125k in profits.
Riz retweeted
ethan asked to be in 404
i checked his trading style and denied his request
fuck around all you want until you find out, but this is certainly not the sway
1/ Meet @0xEthan
He runs an extraction scheme on pumpfun
buy → callout → sell from side wallets
49 tokens. $125k in profits.
Riz retweeted
I told Dora and Nicole, “I ran out of English tokens today 🤣” — and it was true.
Tonight I was at the MetaMask dinner in Singapore. I came last year too. This year was more chill, and the room was more international: more people from the West, friends from Korea, and MetaMask users from many other places, not only the Chinese-speaking community.
The dinner was well run. What stayed with me was their determination to make the product better.
I spent almost an hour and a half with Riz and Moz on exactly that. We went from large questions, how to make the UI and UX simpler, down to small ones, how the PNL chart should separate realized profit from unrealized profit.
Seeing Joseph Lubin there made that commitment concrete. MetaMask is working toward a social trading product that is useful, safe, and competitive. I am looking forward to the official launch.
@ethereumJoseph
Riz retweeted
hello, @MetaMask
Yesterday, I spent four hours talking with team members from various departments at MetaMask. We spent the entire time discussing what we’re currently working on, what needs to be fixed, and what else is needed.
Riz retweeted
JUST IN: Proposal for 'Shielded Bitcoin' brings Tornado Cash-style privacy directly to mainnet.
• Peg BTC into encrypted notes privately
• ZK-proofs hide sender, receiver, amount
• Break on-chain link when pegging out
🫡 Nicholas Gregory @gregory_nico
Riz retweeted
Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead.
You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier intelligence. You’ve also claimed the lead is widening because of recursive self-improvement.
I don’t see what you see in the lab. If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible.
But stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier.
Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack. The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability. Call it alignment if you want. It is also just giving customers what they want.
Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders’ “shut it all down.” China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well.
So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system. So just do it.
If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.
Riz retweeted
Elon predicted 100x gains in AI intelligence at fixed model size. It was at the edge of plausibility when he said it. Turns out it's now a fact. We are nowhere near the ceiling.
The meta-direction at the application layer is clear: buyers increasingly want to pay for outcomes, not seats or usage.
Intercom’s Fin is a strong example: $0.99 when it delivers a support outcome, such as resolving a customer conversation, rather than for every AI interaction.
That shifts more of the performance risk to the vendor, where it belongs.
Most AI apps are pricing at the wrong layer.
Asked how they'd rather be billed, technical AI buyers preferred credits tied to recognizable value over tokens nearly 2 to 1.
Per-token pricing anchors your product to a cost curve that keeps falling.
Full piece from Tugce Erten and Sarah Wang on why customers want legibility, not tokens: a16z.news/p/you-are-not-a-mo…
67
Riz retweeted
🇪🇺BIG MOVE: Revolut is launching its own EURO stablecoin.
EURR will roll out in Denmark, Poland and Portugal before expanding across the EEA later this year.
Revolut already has 80M+ customers globally and 16M+ crypto users, giving the token a massive distribution network from day one.
But euro stablecoins remain around €450M versus roughly $300B for dollar-backed tokens.
EURR, issued by Stripe-owned Bridge, is only the "first step", with more currency-backed stablecoins planned.
Revolut is putting its global currency business on-chain.


































