Ray Xiao

657 posts

Ray Xiao banner
Ray Xiao

Ray Xiao

@_RayXiao

Investing @OKX_Ventures. Curiosity is the North Star. Technology diffusion is as vital as innovation. Prev @IOSGVC Opinions are my own.

Katılım Ekim 2011
490 Takip Edilen1.9K Takipçiler
Sabitlenmiş Tweet
Ray Xiao
Ray Xiao@_RayXiao·
🧵 1/ Beyond the Checkout Page: Who Will Build the Economy for Agentic Commerce? A deep dive into the battle between human-centric design and machine-native protocols. As AI agents evolve from assistants to autonomous entities, they're reshaping commerce. Here is the full link of the research: okx.com/en-sg/learn/be…
OKX Ventures@OKX_Ventures

The future of commerce is here: Agentic Commerce is redefining how we shop, pay, and trust in a machine-driven economy! From AI agents handling purchases to crypto-powered, machine-native payments, we're on the cusp of a revolution. 💸 Read more: okx.com/en-sg/learn/be…

English
6
1
26
5.3K
Ray Xiao
Ray Xiao@_RayXiao·
RFQ vs CLOB can be boiled down to one question: who profits from your trade. Pool LPs take it (socialized broker); external MMs compete for it (TradFi); or skip the internalization middleman entirely, user trades directly against MMs on the public book (DeFi CLOB). CLOB camp says RFQ doesn't provide real onchain price discovery or composability. RFQ camp says CLOB's bootstrapping costs are too high for medium and long tailer RWA assets Both models have their own niches, the endgame is hybrid
OKX Ventures@OKX_Ventures

The onchain RFQ vs CLOB debate dominated CT last week. RFQ in crypto now covers two completely different businesses that happen to share a name. 1. The first is the multi-maker RFQ aggregator. UniswapX, 1inch Fusion, CoW. External fillers, solvers, and MMs compete on private quotes, user takes the best fill. Economically this is what "RFQ" means in TradFi. Uniswap cleared over $1T in 2025 volume (TokenTerminal). 2. The second is a single-counterparty pool with an RFQ-style interface. One internal pool sits on the other side of every trade and hedges offchain. No external maker competition. The business model isn't TradFi RFQ at all, it's Citadel-style retail internalization: Robinhood routes flow to the wholesaler, the wholesaler internalizes, user gets a small price improvement, wholesaler pockets the spread. Two different economic structures. Multi-maker compresses spread through competition. Single-counterparty captures spread through monopoly pricing plus cross-domain hedging. And TradFi RFQ doesn't actually work the way most onchain pitches imply. The dealer's whole business is sitting on balance sheet, and they don't just warehouse and wait for the offsetting RFQ. They internalize against future client flow, offset in the interdealer market, hedge piecewise across correlated assets, whatever the book and the day call for. The whole loop stays off the public book and dealer PnL is compensation for carrying inventory. A maker who hedges back-to-back on a CLOB the moment a quote fills is just running a single-maker CLOB clone. The onchain versions split along these lines too. Multi-maker inherits the competitive pricing but most makers hedge immediately rather than warehouse. Single-counterparty is its own thing entirely: retail internalization, ported onchain.

English
0
0
3
236
Ray Xiao
Ray Xiao@_RayXiao·
hey we are here now
Ray Xiao tweet media
English
1
0
1
66
stepan
stepan@cyntro_py·
Why AI-native startups and founders specifically? LLMs as a technology that lets you scale intelligence across an organization. To do work that requires attention, reasoning, logic, and decision-making many times cheaper and faster. The first and most obvious way to apply this tech is “hey, let’s automate something!” Call prep, writing reports, reviewing contracts, managing tasks, analyzing data. This sells — especially to the people who control the budgets for it. They’re simple folks: they see that swapping manual hours for tokens saves $100K a year on a single task and costs $50K — let’s go. Automation is easy to sell to anyone with business experience who, most likely, runs one. But this kind of optimization misses the most important question: why are we automating a process that shouldn’t exist in the first place? The dramatic drop in the cost of intelligence, along with shifting user and employee behavior, means a lot of legacy processes simply become unnecessary. Why do you need reports and dashboards if every employee has an agent that can go straight to the database and ask the question it needs? Or, even simpler, constantly monitor it and proactively flag what to pay attention to and when? This is exactly where the AI-native innovator’s dilemma kicks in. In an existing company it’s easy to start automating, but it’s extremely hard to sell — let alone actually pull off — a change in org design, change management, the reinvention of processes. In a startup you can build them the new way from day one: - a company that learns about itself and cybernetically improves - all information and decisions tokenized from day one - every decision analyzed by an agent by default and escalated when needed and so on, there are hundreds of these changes, many of which we’re studying in the Monastery
cyber•Fund@cyberfund

Introducing the Monastery for AI-native founders. A single builder can now outperform a publicly traded company. $2 million. 12 weeks. Do the impossible.

English
3
1
29
3.6K
Ray Xiao retweetledi
Axis
Axis@AxisFDN·
18.2% net APY. $32.5m TVL. 6 weeks of private beta. The latest Axis tear sheet - published directly to the timeline for everyone, not just LPs. Uncorrelated to the market. Delta-neutral, synchronizing prices globally. Full report in thread.
English
7
2
47
10.2K
Ray Xiao
Ray Xiao@_RayXiao·
Huge respect to C4 @code4rena @sockdrawermoney and the whole security researcher community. I still remember one of my favorite panels, people are debating about the community driven security model. Spicy, but also love that energy we had. youtube.com/watch?v=_Yul_f…
YouTube video
YouTube
Code4rena@code4rena

After careful consideration, we’ve made the decision to wind down @code4rena. This community has meant a great deal to everyone who has been part of building it, and sharing this news is not easy.

English
0
0
2
140
Ray Xiao
Ray Xiao@_RayXiao·
We've been tracking the AI commerce shift at OKX since early last year and wrote about an article about how we believe agent commerce will rewrite the foundations of business: okx.com/en-sg/learn/be… The biggest shift in commerce is moving from the checkout page to the intent layer: from clicks to intents, from executors to delegators. The closest point to the customer is no longer the storefront or the app. It’s the AI agent that holds the user’s intent. Value will accrue not to whoever owns the traffic, but to whoever owns the protocol agents transact on. Card rails were built for humans. Open protocols, native wallets, and programmable rules are the stack built for machines. This is where crypto stops being an alternative. We believe it will become the default substrate for agentic commerce. APP (Agent Payments Protocol) is our big move in shipping the thesis. It enables agents to: • Discover and engage with each other • Agree on terms and pricing • Execute payments across different models (one-time, streaming, escrow) • Complete and settle work end-to-end without constant human coordination In that sense, this is less about “payments” in isolation, and more about infrastructure for agentic commerce. Just as open protocols shaped the early internet, we think similar primitives are needed for an AI-native economy. Read white paper here: web3.okx.com/whitepaper/okx…
OKX@okx

x.com/i/article/2049…

English
0
0
2
171
Ray Xiao retweetledi
Tushar Jain
Tushar Jain@tushar_jain·
0/ DeFi needs circuit breakers and other safety mechanisms which slow down large transactions and provide time for reaction. Borrow lend protocols should not allow a new user to show up with a $300M position and take out a loan against it immediately. Some ideas:
English
42
18
285
69.8K
Ray Xiao retweetledi
Dan Robinson
Dan Robinson@danrobinson·
I'll have a lot more to say about our learnings from the hackathon, but the clearest takeaway is that on optimization problems, agents beat experts I designed the prediction market challenge myself and crafted a solution using domain knowledge But my score got absolutely crushed by people who didn't even look at the puzzle x.com/ryanli/status/…
English
4
5
50
6.5K
Ray Xiao retweetledi
Etherealize
Etherealize@Etherealize_io·
Marc Andreessen: “This is the grand unification of AI and crypto” “I think AI is the killer crypto app… It’s now obvious that AI agents are going to need money. It’s already happening.” Marc explains: “My friends, who are the most aggressive users of OpenClaw, have given their Claws bank accounts and credit cards. And not only have they done it, but it’s obvious that they needed to do it… It’s just completely obvious. The number of people who have done that today is, I don’t know, probably 5,000 or or something. But it will grow. That’s how these things start.” Source: @latentspacepod (Apr 2026)
Etherealize@Etherealize_io

x.com/i/article/2042…

English
81
233
1.8K
549.8K
Ray Xiao
Ray Xiao@_RayXiao·
Had the pleasure of hosting this panel with four solid builders @moonshot6666 @dex_chen_V @likeAScientist @PiP_Quant who are building AI products for prediction markets and crypto trading. Rare to get this many sharp, honest, first hand perspectives in one room — on everything from why LLMs are exclusively backwards looking devices that need world models and causal reasoning to be useful, to why the real job of AI in trading isn't generating alpha but automating the grunt work so retail traders can operate at 70-80% of a professional's level, to how the products we build today for humans will soon be serving their agents instead. One of the more valuable conversations I've been part of on AI × crypto trading.
OKX Ventures@OKX_Ventures

x.com/i/article/2038…

English
1
0
8
316
Ray Xiao retweetledi
Semantic Layer
Semantic Layer@SemanticLayer·
Why LLMs are a natural fit for prediction markets? These markets are driven by massive continual news flows: politics, culture, economics, sports. LLMs are very good at compressing that flood into something usable. @dex_chen_V on why that matters
English
0
3
12
810