←
knwang4,002 posts
knwang avatar

knwang

@knwang

intention is all you need // building @khalani_network // cofounder @nervosnetwork // dm open

◷ Joined June 2008▧ 35 Photos and videos
knwang avatar
knwang@knwang·Jun 27

agents don't get better; their knowledge layer and context injection does. building knowledge layer is all about normalization, denormalization and relationships between pages. we've been working on this for our internal agents and it's come to a point that we're basically define relational graphs in natural languauges, not very different from databases.

knwang avatar
knwang@knwang·Jun 26

increasingly feel skills is an anti-pattern. objective based evaluation loops are much better.. define standing objectives, inputs and let agents define quality standards on delivery and let agents run in loops.

knwang avatar
knwang@knwang·Jun 5

stablecoins will clear workflows, payments are just embedded executions.

Prediction: by 2028 most stablecoin volume is B2B treasury, not retail remittance. Remittance gets the headlines -- it's sympathetic and visible. But the dollars are in supplier payments, intercompany sweeps, and payroll. Boring, recurring, enormous. The boring flows always win on volume.

knwang avatar
knwang@knwang·Jun 5

Great writeup by Jessy. Any bilateral execution-focused use cases likely won't need blockchains; multilateral execution use cases only need blockchains when value movement itself is the use case. Excited on what you guys are building next!

Jessy Huang @goodhartproof Jun 3 Article a year inside agentic payments: the uncomfortable truth Thanks to @austingriffith, @dongossen, @ekrahm, @programmer, @RosuGrigore, @0xfishylosopher and @JackSimison for the review and feedback. I've spent the last year building infrastructure for the agent

knwang avatar
knwang@knwang·Mar 19

> I think that agentic payments are the killer app state channels never had. 👍👍👍 off-chain replicated state machines with superseding signed state that use blockchain for escrow and settlement. They’ll come for payments, and stay for programmable coordination.

Liam Horne @liamihorne Mar 18 Article State Channels and Agentic Payments State Channels and Agentic Payments I’ve spent almost a decade obsessed with one idea: software paying software. Back in 2017, a bunch of us poured years into building state channels on Ethereum,

knwang avatar
knwang@knwang·Mar 7

Claude code re-discovered Unix, and taught us that powerful coordination doesn’t require rich, bespoke standards, but a small set of simple and universal primitives: - cli powers everything - everthing is a file - every task is pipe-able Simple interfaces win. Unix won. Git won. HTTP won. REST won. But almost all current agent protocols are designed like ESB and CORBA. It gets worse for agentic payment, commerce or transaction standards. They won’t work, not because agents don’t need standards, but because they’re trying to standardize the wrong layer. We keep trying to standardize rich semantics, shared ontologies, negotiation flows, and high-level message schemas. That is exactly where heterogeneity is highest and agreement is hardest. Unix did not win by standardizing meaning. It won by standardizing composition. Don’t standardize what agents think. Standardize how value moves when they agree.

499 Group avatar
⟳ knwang retweeted
499 Group@499_Group·Feb 25

💡Patrick Young @patkyoung — Head of Growth, Khalani @khalani_network "Agentic payments need bulletproof execution rails." 🤖 • The bottleneck isn’t AI intelligence — it’s coordination, verification, and deterministic settlement. • Without trust-minimized execution, AI is just another API. • Agents need escrowed capital, solver accountability, and protocol-level guarantees. • The real blocker today? Fragmented chains, liquidity, and execution venues. • Autonomous agents require unified intent routing + cross-chain liquidity. The future isn’t on-chain AI. It’s verifiable, economically-enforced execution.

knwang avatar
knwang@knwang·Feb 25

with our very own @patkyoung !

Panel Highlight | How AI Becomes Trust-Minimized at Scale Hosted by @499_DAO × @0G_labs × @hetu_protocol At The Scaling Summit @ ETHDenver, leaders: Michael Dressler @mdressler24 — Head of Success, 0G @0G_labs Patrick Young @patkyoung — Head of Growth, Khalani @khalani_network Michael Berman @mikebermantoo — Co-CEO, Tectonic Labs @tectonicxyz Vinay Kumar — Co-Founder, Flashback Labs @FlashbackLabs Explored what it really takes for AI-native infrastructure to become composable, verifiable, and production-ready. AI doesn’t fail because of model capability. It fails when execution and settlement aren’t trust-minimized. #TheScalingSummit #AIInfrastructure #DeAI #Web3AI #ETHDenver Watch the discussion ↓

3:53
knwang avatar
knwang@knwang·Feb 6

dangerously skip permissions dangerously skip permissions dangerously skip permissions

knwang avatar
knwang@knwang·5 Dec 2025

the combination of private, general purpose and high throughput would be extremely compelling

Ali Yahya@alive_5 Dec 2025

Privacy will be the most important moat in crypto. Why? Because secrets are hard to migrate. Everyone is launching a new "high performance" blockchain lately. But these chains are hardly different from one another. Blockspace is functionally the same everywhere. And with bridges that make moving between chains easy, that blockspace is now accessible *from* everywhere. Mercenary users and capital quickly arriving on a chain to farm an airdrop can leave just as quickly to farm the next one on another chain. The reality is that if your "general purpose" chain doesn't already have a thriving ecosystem, a killer application, or an unfair distribution advantage, there's very little reason for anyone to use it or build on top of it. Performance alone is no longer enough. Privacy is the one feature that everyone agrees is critical for the world’s finance to move onchain. It’s also the one feature that almost every blockchain that exists today completely lacks. For most chains, it has been little more than an afterthought until now. Privacy by itself is sufficiently compelling to differentiate a new chain from all the rest. But it also does something more important: it creates chain lock-in. Bridging tokens is easy, but bridging secrets is hard. As long as everything is public, it's trivial to move from one chain to another, thanks to bridging protocols like LayerZero. But, as soon as you make things private, that is no longer true. There is always a risk when moving in or out of a private zone that people who are watching the chain, mempool, or network traffic will be able to figure out who you are. Crossing the boundary between a private chain and a public one—or even between two private chains—leaks all kinds of metadata like transaction timing and size correlations that makes it easier to track you. Compared to the many undifferentiated new chains whose fees will likely be driven down to zero by competition, blockchains with privacy have a much stronger network effect. When you're on public blockchains, it's easy to transact with users on other chains—it doesn't matter which chain you join. When you're on private blockchains, on the other hand, the chain you choose matters much more because, once you join one, you're less likely to move and risk being exposed. This will create a winner-take-most dynamic. And because privacy is essential for most real-world use cases, a handful of privacy chains will own most of crypto.

knwang avatar
knwang@knwang·11 Nov 2025

the challenge is the opposing force of verifiability, agency and interoperability - direct verification over agent's computation lowers agency and less future-proof - optimistic verification allows higher agency but increases agents operating cost with bonds - in either situation, since delegation is over time and verification is not atomic with every agent action, interoperability between risk policy curators, markets and agents is not trivial

knwang@knwang10 Nov 2025

It's one thing to use x402 to receive one-time use for a service, but entirely another to source over-time capital delegation to fund the working capital of a service with pre-agreed economic terms. It's like comparing an off-chain verifiable agent that can submits zk proofs for its execution vs. a zk-rollup. As the demand for the former increases, the demand for the latter will follow too for any blockchain adjacent services.

knwang avatar
knwang@knwang·10 Nov 2025

It's one thing to use x402 to receive one-time use for a service, but entirely another to source over-time capital delegation to fund the working capital of a service with pre-agreed economic terms. It's like comparing an off-chain verifiable agent that can submits zk proofs for its execution vs. a zk-rollup. As the demand for the former increases, the demand for the latter will follow too for any blockchain adjacent services.