Thea Høgholm

Editor @BeaconLayerCast | retweet ≠ endorse | Tips: [email protected]

[Thea Høgholm @TheaHogholm]

the great news is that pretty much all of this is totally preventable

ethereum and ETH are rapidly growing to global ubiquity. this doesn't change that one bit

future defi solutions won't make these amateur errors (they'll eventually seem like amateur errors... already do to many experts)

attacker supremacy from AI won't last, it's working through a backlog of exploits. the limit result will be much permanently more secure protocols and eth's neutrality and capital gravity well will be more valuable than ever

if you have ETH lent in aave, imo you should get out right now by using onchain limit orders to sell to whale loopers who are actively buying aETH unwind leverage. you can take as little as a 1.1% haircut based on ongoing txns in mid 6-figs, which is much lower than some estimates of final socialized losses.

why can ETH lenders take only a 1.1% haircut now? because loopers want to avoid liquidation from high rates caused by 100% utilization. on the bad debt side, loopers actually win from socialized losses because their debt token (aETH) is the one taking the haircut, so they'd owe less in ETH terms.

the most remarkable thing in this crisis is clearly that billions of dollars in backbone eth lending on aave were in fact exposed to signer risk in a 3rd party bridge... effectively some random downstream fellow was actually an aave admin.

aave additionally has negligently low borrow rates during 100% utilization, leading to extremely dangerous illiquidity. what if ethusd crashed for any reason... eg. if stocks were open and a politician said the wrong thing, btc goes down 5%, eth goes down 8%... this can lead to broader contagion and bad debt.

protocols and their teams like fluid (who've had low level dynamic withdrawal rate limits in protocol from day 1 so can't be insta drained) and spark (who seem to have excellent scientific gov and no exposure for eth lenders to 3rd party bridge admins) deserve respect and attention for doing what they knew was right and possible even before the ecosystem had a forcing function to care about it. same goes with other kinds of security practices that are still fringe, maybe including formal verification and ipfs hashes for frontends

nearly 24h since the attack, the lack of material updates from affected protocols, including kelp, layerzero, and aave, suggests to me the ongoing severity of the situation. many factors are in play, there's probably no great solution, somebody is going to lose big

is just the bridged rsETH (that argubly took bridge risk intentionally) fully on the hook for the bridge failure, and L1 rsETH should be unaffected? however L1 rsETH *was* affected due to gov choices in aave. does aave's junior debt program, umbrella, take the full wipeout? however umbrella's terms & conditions say they have no bridge risk, which aave gov effectively violated without umbrella holders realizing it. does layerzero bear responsibility for allowing their users to be subject to terrible admin config in one of their ecosystem bridges? i'm probably missing aspects here, it's very messy.

Just Use Aave is dead... nobody is going to Just Use Anything anymore

the future of this industry is to do the smart obvious stuff even when it's unpopular, like withdrawal rate limits, better interest rate gov, avoiding toxic market share steroids like degen bridge looping affordances.

and for 10x better user recognition and higher standards around protocol hygiene and security differentiation. degen stuff is fun and amazing but only when you understand the true risks

in sum, if you are lending ETH in aave, get out now at a ~1.1-1.5% haircut by selling to loopers actively unwinding because when the dust settles, a material haircut for Aave v3 ETH lenders is a possibility

ethereum and ETH are growing well to global ubiquity and will be massive net beneficiaries of our industry successfully navigating this crisis season of backlogs of exploits discovered by AI and preventably poor practices in defi architecture/gov. trillions await

[BeaconLayer Podcast @BeaconLayerCast]

“What if AI gives us a once-in-a-generation chance to rebuild the internet from the ground up?”

@kenzimori sits down with Karan Sirdesai @karansirdesai, Founder of @miranetwork, to unpack the founder journey behind Mira and the early conviction that AI would become far more than another tech cycle.

They trace Karan’s path from teaching himself to build, sending cold DMs, experimenting with crypto arbitrage, and working alongside figures like Balaji Srinivasan and Sandeep Nailwal, to spotting the AI shift before it became obvious to the broader market.

The conversation also explores why Karan has consistently chosen the unconventional route, how his time at Accel exposed him to frontier AI companies early, and why the pace of AI progress made one thing clear: this wouldn’t just change startups or software, it would reshape how humans work, think, create, and interact with the internet itself.

[BeaconLayer Podcast @BeaconLayerCast]

“What if the most important AI infrastructure isn’t the biggest model — but the one you can actually trust?”

@sachimiyasaki sits down with Karan Sirdesai @karansirdesai, Founder of @miranetwork, to explore how Mira is approaching decentralized AI from a more focused angle: making machine intelligence reliable enough for real-world use.

They discuss why Mira is not trying to become a full-stack AI protocol, how its approach differs from networks like Bittensor, Ritual, and Sahara, and why focusing on trust and verification could make Mira a stronger fit inside the wider AI infrastructure stack.

The episode also breaks down Mira’s reliability architecture, from its core verification layer to node-level consensus, and why the next evolution of AI APIs may not be about simply accessing more models, but about knowing whether the answers you receive can actually be trusted.

[BeaconLayer Podcast @BeaconLayerCast]

“What happens when AI stops being a tool and becomes the operating layer of the internet?”

@dikshawells sits down with Karan Sirdesai @karansirdesai, Founder of @miranetwork, to explore a future where AI is no longer just answering questions, but taking action, making decisions, and handling large parts of our cognitive workload.

They discuss the upside of this shift: the rise of “infinite hires,” a world where most knowledge work can be automated or amplified, and a productivity wave that could completely reshape how companies, teams, and individuals operate.

But the conversation also gets into the darker side of that future. As AI becomes more capable, the risks grow with it: blind dependence, high-stakes failures, malicious autonomous agents, and the possibility of humans sharing the world with a new kind of intelligence we are not fully prepared to control.

[BeaconLayer Podcast @BeaconLayerCast]

This week on the podcast, we’re joined by Karan Sirdesai @karansirdesai, Co-Founder and CEO of Mira Network @miranetwork.

We cover Karan’s path from his university days and early crypto experiments to building Mira, an infrastructure layer focused on making AI more reliable, verifiable, and safe to use at scale.

The episode goes deep into one of AI’s biggest unsolved problems: hallucinations. Karan breaks down why unreliable outputs are such a major blocker for real-world adoption, especially in high-stakes areas like finance, healthcare, and other trust-sensitive industries.

We also discuss the early “aha” moment behind Mira, shaped by experiments with GPU rentals, AI workflows, and the insight that multiple models could work together to verify outputs through consensus.

Karan also shares how working with Balaji Srinivasan, his unconventional founder journey, and Mira’s AI + crypto-native team influenced the company’s mission: building decentralized infrastructure that helps make artificial intelligence more trustworthy, dependable, and useful in the real world.

[BeaconLayer Podcast @BeaconLayerCast]

After 7 years, Aztec’s Ignition mainnet is live.

Yet zero transactions or apps work yet. The chain is deliberately empty – because true protocol-level privacy can’t be rushed.

Here’s how this phased, decentralization-first launch positions Aztec as the leading private L2 on ETH👇

~~ Analysis by @punk1831 ~~

What's Actually Running

Think of Ignition like Ethereum's beacon chain from 2020. The governance and consensus infrastructure is operational, but the execution layer remains offline. The team is running what amounts to a live stress test with real money on the line.

Each sequencer staked at least 200K $AZTEC tokens to participate. They're producing blocks, provers are generating validity proofs, and the whole system is settling on Ethereum, just without any transactions.
The goal of running Ignition with real economics for 2-3 months will (hopefully) surface any remaining issues before transactions go live in early 2026, while setting the network up to be decentralized from day one.

The Decentralization Push

In Aztec's eyes, launching an L2 with a centralized sequencer from the get go rarely translates to decentralization down the road.

Centralized sequencers generate $40-150M annually in fees. Once you're locked into those cash flows, decentralization means making transactions slower and more expensive. The tension never resolves.

Instead, @aztecnetwork will launch fully decentralized from day one across three dimensions:

➢ Ownership is decentralized through $AZTEC token holders who control network parameters, fee schedules, and protocol upgrades.

➢ Block Production runs through 617 decentralized sequencer nodes using proof-of-stake. These nodes order transactions and produce blocks. To prevent any single party from gaining control, a small committee of sequencers is randomly selected to validate blocks before they are submitted to Ethereum.

➢ Proving is permissionless from the get-go. Provers generate the zero-knowledge proofs that cryptographically confirm all transactions in a batch are valid. They aggregate blocks and submit a single, final proof to Ethereum for verification, guaranteeing the integrity of the entire rollup.

When transactions go live, Aztec will qualify as a Stage 2 rollup, the highest decentralization tier for L2s. Most chains have pushed boundaries in one direction. Hitting all three pillars simultaneously is rare.

In Aztec's eyes, decentralization isn't optional for privacy. Centralized sequencers would face pressure from governments to install backdoors. Privacy requires cryptography plus decentralization, not one or the other.

What Happens Next

There are two major upcoming events, one technical and one token-related.

On the technical side, Ignition will remain live for 2-3 more months with sequencers producing empty blocks while the team monitors for issues. Early 2026 is when transactions flip on. Users will be able to send payments, deploy smart contracts, and interact with applications. By the end of 2026, block times should drop from the current 36-72 seconds down to 4 seconds, faster than Ethereum's 12-second blocks.

On the token side, the pre-allocation for the $AZTEC token sale is currently live, with the sale beginning December 2nd and running for 4 days. The sale uses @Uniswap's continuous clearing auction mechanism, meaning if you bid early, part of your bid clears at early prices and part clears later. This levels the playing field between early and late participants while letting price discovery happen naturally. When the auction ends, it automatically creates a Uniswap V4 liquidity pool at the final clearing price.

To participate in the sale, you must register prior to December 2nd.

For compliance, Aztec is using @ZKPassport, enabling people to prove cryptographically that they're from allowed jurisdictions and not on sanctions lists without traditional KYC. The sale is open to US retail and nearly every country worldwide, with the exception of sanctioned countries on the standard OFAC list.

The current 500 sequencers already staked $AZTEC tokens they purchased in a whitelisted genesis sale. They're earning rewards in $AZTEC right now. However, all tokens, whether from the genesis sale, the current public auction, or insider allocations, are non-transferable until Token Generation Event (TGE).

There is no set date for when TGE occurs, rather the community votes on it. However, the earliest date it can go live is February 11th, 2026. Once TGE happens, tokens purchased in the public auction unlock 100%.

7 Years in the Making

Overall, Ignition and the $AZTEC token sale demonstrate both the complexity of successfully executing privacy, as well as the extent to which Aztec is going to get this right.

First you have the need for decentralization from the get-go to ensure privacy endures, a feat unaccomplished by countless L2s launched so far. Then you have the tension between privacy and compliance, which the token sale's integration with ZK Passport helps solve.

Regardless of how mainnet goes, and I'm hopeful all goes well, this launch process shines as a testament to diligent design, demonstrating that forces like decentralization, privacy, and compliance can all coexist

[BeaconLayer Podcast @BeaconLayerCast]

“Gary Gensler was teaching a class while I was on campus. I skipped it—and I’ve regretted it ever since.”

@kenzimori links up with @annakaz, founder of @Vana, about those small early moments that quietly shape how you think—like brushing up against future regulators, and realizing way too late which conversations you should’ve been in the room for.

[BeaconLayer Podcast @BeaconLayerCast]

“Back then I was nerding out on central banking—then I showed up at MIT and ended up mining Ethereum from my dorm room.”

@kenzimori catches up with @annakaz (@Vana) about her early arc: how a straight-line interest in institutions and monetary policy flipped into crypto after she found the small, ideology-heavy MIT Bitcoin crowd—right as Ethereum was starting to take off.

[BeaconLayer Podcast @BeaconLayerCast]

The pace of decentralized AI is becoming a builder’s advantage — better tools, faster feedback loops, and entirely new capabilities emerging in real time.

But that speed comes with a cost: it’s hard to ever be fully offline when the frontier moves this fast and yesterday’s assumptions can age out overnight.

@dikshawells sits down with @annakaz, founder of @vana, to unpack the whiplash speed of decentralized AI.

[BeaconLayer Podcast @BeaconLayerCast]

This week’s episode features Anna Kazlauskas (@annakaz), founder of @Vana.

We talk about her jump from traditional finance into decentralized AI—and what it takes to build a platform where people own their data and choose how it trains models. We also dig into tokenomics as core infrastructure, not a side quest.

We cover the real startup stuff too: early hires, mission fit over resumes, and staying focused when the narrative shifts weekly. And we close on the AI data crunch—why private data is uniquely valuable, how data DAOs can unlock it, and what separates DAOs from trusts or unions.

[BeaconLayer Podcast @BeaconLayerCast]

Last week, Ethereum’s ecosystem (L1 plus rollups) blasted through a new high, briefly hitting 24,000 transactions per second.

From 0.7 TPS in 2015 to regular spikes above 10,000 TPS today on @growthepie_eth, the modular bet is delivering explosive gains after a decade of building.

Welcome to Ethereum’s exponential age.👇

~~ Analysis by @kenzimori ~~

To be sure, the bulk of this current TPS surge is stemming from Lighter, the newer perps L2 whose custom appchain architecture minimizes what data touches Ethereum.

@Lighter_xyz just posts compressed state diffs and proofs to the L1 while keeping its high-frequency order flow offchain. This zk appchain design is unique in the rollups scene today, but more teams will experiment with this model and extend it in new directions.

Beyond this design evolution, Ethereum's roadmap has plenty of ecosystem-wide advances on the way that will help push performance gains. Foremost to mind is PeerDAS, which the Fusaka upgrade will bring to mainnet next month.

PeerDAS will be a powerful upgrade, as it's projected to facilitate around an 8x increase in Ethereum's blob capacity. With improved data availability, rollups are set to march past 1 million in ecosystem TPS in short order.

For instance, @base hit 1,500 TPS in June 2025 with Ethereum's current blob limits. Blob capacity going up 8x makes 10,000+ TPS feasible for the L2 at some point next year.

This math applies to zk appchains like Lighter, too. If Lighter can handle ~45,000 TPS today, it can potentially pass 350,000 TPS in 2026. Of course, there will be impactful project-level advances as well. ZKsync's upcoming Atlas upgrade has the potential to facilitate 15,000+ TPS for ZK Stack L2s. And that's just one stack and one upgrade.

So yes, Ethereum is scaling horizontally, and the prospects here are impressive. But Ethereum also has considerable vertical scaling potential. There are ongoing efforts, like EIP-7938 and "Lean Ethereum," that can help the L1 reach 10,000 TPS in its own right.

With this "all of the above" approach, we can dream big. We can build a new substrate for all the world's commerce and culture. And all of that builder potential is possible precisely because Ethereum is going tall and wide in its scaling.

This is the endgame: many chains spreading out to the horizon in every direction for any need, all anchored around an incredibly secure and robust network that's worthy of powering an entire civilization.

The progress here is clear. Meanwhile, the Ethereum community will continue to create its own destiny, just as it always has. We know the path forward, and nothing can stop us now.

[BeaconLayer Podcast @BeaconLayerCast]

It's been one month since Hyperliquid's HIP-3 went live, letting anyone stake 500K $HYPE can now launch custom markets backed by the platform’s deep liquidity.

The result is an exchange where you can long or short anything: stocks via Trade or Felix, commodities, bonds via Aura, pre-IPOs via Ventuals, even Pokémon cards via Trove.

Learn how HIP-3 works and how it make impact Hyperliquid 👇

~~ Analysis by @dikshawells ~~

The upgrade works like this: a deployer stakes 500K
$HYPE (~$19.3M at time of writing). They can then list three markets for free before entering an auction process to secure additional slots.

For each market they launch, the deployer sets leverage limits, configures the oracle, and manages key technicalities. To ensure acceptable standards, deployers risk having their stake slashed, though Hyperliquid notes this mechanism is temporary and expected to fade as standards and tooling improve.

Once live, the deployer earns 50% of the fees from their markets, with Hyperliquid taking the other half. To balance revenue, HIP-3 market fees are set at double those of standard markets, keeping @HyperliquidX's take roughly equivalent.

While most deployers are still building, early activity from just one HIP-3 market already live — to the tune of $1.3B in volume — paints a positive picture that the upgrade's potential may match its hype.

What Will the Impact of HIP-3 Be?

As a result of how it's designed, HIP-3 introduces new supply crunches on $HYPE, additional revenue for buybacks, and potentially increases rewards earned by stakers and traders.

➢ Locking up $HYPE: Each deployer must stake 500K
$HYPE, effectively removing that amount from circulation. The result is persistent buying pressure as new deployers acquire $HYPE to secure their slots. For example, Trove raised $20M to purchase $HYPE for its launch. Further, Hyperliquid Digital Asset Treasuries (DATs) like @HyperionDeFi and @HypeStrat have already begun exploring how to get involved in HIP-3, alleviating the threat of these vehicles dumping their tokens as we're seeing more DATs do.

➢ Additional revenue for buybacks: The 50/50 fee split on HIP-3 markets provides a new inflow to the protocol Assistance Fund, which uses 97% of all fees to buyback its token. Because HIP-3 market fees are set higher than standard ones, this stream will not be reduced by the split in fees with the deployer, potentially offering a significant source for $HYPE buybacks if even a handful of markets achieve sustained volume.

➢ Incentive Wars: A likely next phase is direct competition among deployers for trader flow, especially given the success of @tradexyz's XYZ100 HIP-3 market, which generated $100K in fees before it even reached two weeks. Expect escalating incentive programs — liquidity mining, fee rebates, staking boosts — as providers fight to draw and retain users. These will likely extend to $HYPE stakers too as validators vie for stake to participate in secondary economics like "exchange-as-a-service" models, where staking providers like @kinetiq_xyz
essentially crowdsource $HYPE to lower the cost of launching a market.

Together, these dynamics tighten HYPE's supply, expand its buyback base, and create new competitive layers across the ecosystem.

How Could HIP-3 Fail?

HIP-3's success will depend on two things: quality markets launching, and those markets generating sustained demand.

Permissionless listings don't guarantee quality. A HIP-3 market is only as strong as its deployer — how they configure leverage, oracles, and risk parameters. Deploying non-crypto or thinly traded assets like stocks or bonds requires continuous data and stable pricing. Without that, markets face thin liquidity, wide spreads, and erratic execution that will quickly drive traders away.

Oracle providers like @redstone_defi are building hybrid systems that blend onchain and offchain data, maintaining live pricing even when the base asset isn't trading. HIP-3's architecture allows deployers to implement proper oracles into individual markets and tailor risk parameters accordingly.

But demand remains the harder part. As @felixprotocol's founder Charlie (@0xBroze) notes, the lion's share of Hyperliquid's volume comes from five markets, mostly composed of major assets like $BTC, $ETH, and $SOL. Smaller assets tend to be left with little natural flow, meaning nascent, niche assets launched via HIP-3 will face a cold-start problem. Without early liquidity, traders hesitate; without traders, liquidity providers leave.

If simply introducing novel markets isn't enough to spark activity, deployers will need to experiment with market structures and pairs, introducing new collateral for perps or unique pair-markets like $BTC / $GOLD. Incentive programs should help smooth the initial launch, but in the end, these markets will have to stand on their own.

Ultimately, HIP-3's trajectory depends on the competence of its deployers. The framework is in place, but its outcome will hinge on whether deployers can build markets that trade well and sustain activity.

Final Thoughts

HIP-3 represents another structural bet on decentralization — a next step for Hyperliquid shifting responsibility for growth from the protocol to its participants.

Whether it succeeds will come down to the quality of the markets that launch, the liquidity they attract, and the flywheel effects that follow. If deployers can navigate those early hurdles, HIP-3 could define the next phase of onchain market design.

It doesn't need scale in the traditional sense to succeed. As Charlie noted, just a few high-performing markets could validate the model and materially impact both Hyperliquid's growth and $HYPE's price, with one firm,
@FalconXGlobal, estimating $.8B in additional fees if HIP-3 captures less than one percent of Mag7 derivatives trading.

For the platform that keeps defying expectations, rising from a fully-bootstrapped team to become a protocol responsible for earning 35% of all blockchain revenue some months, the success of HIP-3 wouldn't be something I bet against.

[BeaconLayer Podcast @BeaconLayerCast]

Before “crypto AI” became a category, David was already exploring the ideas behind it.

@kenzimori catches up with @david_enim on his early path into crypto x AI — from studying Bitcoin’s economics and computer science during his PhD, to witnessing the 2017 wave of crypto builders up close.

The conversation traces how David’s work across game theory, machine learning, and agent systems eventually shaped his thinking around autonomous agents and open-source coordination infrastructure.

[BeaconLayer Podcast @BeaconLayerCast]

What does it really mean to co-own AI?

@sachimiyasaki sits down with @david_enim, core contributor to OLAS, to unpack how users can move beyond simply using AI — and start participating in it.

With OLAS, users can hold tokens, run agents, or even operate them as businesses. Across DeFi, prediction markets, and other onchain use cases, agents can generate value, while operators share in the upside.

[BeaconLayer Podcast @BeaconLayerCast]

Our host @kenzimori connects with @david_enim, Co-Founder of Valory, to explore what it really takes to build autonomous agents across crypto.

After years at https://t.co/HJjIjVjvMm, one lesson became clear: chains can create focus, but they also create friction. Instead of scaling agents, you often end up selling blockspace, ecosystems, and infrastructure narratives.

[BeaconLayer Podcast @BeaconLayerCast]

“Honestly, I think we’re just a few months away.”

@dikshawells sits down with @david_enim to unpack how close AI agents really are to mainstream adoption — not just for builders, but for beginners and everyday DeFi users too.

After years spent deep in technical papers on agent systems, and now building in the space full-time, David explains why the timeline is compressing so quickly.

And why the biggest challenge may no longer be the tech itself, but getting people aligned around how to actually use it

[BeaconLayer Podcast @BeaconLayerCast]

This week’s episode features David Minarsch (@david_enim), Co-Founder & CEO of Valory.

We dive into David’s journey across crypto and AI, from his early work with https://t.co/SV4qEWB8aW to building Valory and OLAS, and unpack what “autonomy” really means once agents move beyond demos and begin operating in real markets.

The conversation explores the rise of autonomous agents: what they can already do today, where they are becoming genuinely useful, and what breaks when you deploy them inside permissionless, adversarial environments like Web3.

We also dig into the hard problems at the intersection of Web3 and AI: coordination, incentives, security, reliability, and why open-source infrastructure matters if agents are going to become first-class participants in the future of the internet.

[BeaconLayer Podcast @BeaconLayerCast]

Artists mint across chains and platforms, making it tough to showcase work cohesively.

Collectors juggle holdings over Ethereum, Solana, Tezos, and more—wallets scattered everywhere.

That's why Raster, a project tackling this mess, has my full attention.👇

~~ Analysis by @TheaHoegholm ~~

Collecting, not archeology

Introduced in September 2025 by @thefunnyguysNFT and @0xmetaclass, Raster is a new digital art marketplace. It consolidates:

➢ Chains — The platform launched with support for NFTs on Ethereum, Ethereum L2s, and Tezos, with Bitcoin and Solana integrations incoming.

➢ Markets — @raster_art aggregates listings, offers, and etc. from OpenSea, Magic Eden, and beyond. If you list your NFTs on Raster itself, the platform cross-posts to external marketplaces for added reach.

➢ Profiles — A Raster profile compiles an artist's works from all supported chains into a clean, singular gallery. Conversely, collector profiles pull together holdings from any connected wallets, displaying them by artist and with live activity feeds of your recent sales, offers, etc.

For example, one of my favorite artists is Gremplin. When I go to his Raster profile, the page surfaces +40 collections—it's easy to look through and discover pieces that are hard to find elsewhere.

And when I go to my collector profile, I find all my holdings organized neatly by artist.

The combination of cross-chain support and clean, easy-to-navigate UX makes Raster an awesome new resource for onchain creatives and collectors. Royalties are honored here, too.

Of course, the platform only just arrived, so its indexing isn't comprehensive yet. But with 16M tokens and 100k artists tracked so far, the scope is already expansive.

For instance, I noticed Raster indexed an old art collection of mine that I minted on Tezos in 2021. I forgot about the series, and it's basically unknown to anyone else. So it's cool to see Raster going both deep and wide in its coverage.

Things will get deeper and wider too, especially once Bitcoin and Solana support are added. In the meantime, the Raster team's fielding feedback, so if you notice any works or artists are missing from the site, you can use the contact pop-up in the FAQ to let them know!

[BeaconLayer Podcast @BeaconLayerCast]

Last week you might have seen Jesse's post about an agent called Oracle by Noice.

It read: if you like this tweet, you buy Oracle. Whether it worked for you isn't important — it's what Oracle envisions that you should pay attention to.

But first let's understand exactly what Oracle is and how it works.

~~ Analysis by @punk3626 ~~

Developed by the @noicedotso team (the same group behind the viral tipping app on Farcaster), the oracle @noiceagent is a new solution for buying predetermined amounts of whitelisted Internet Capital Markets (ICM) tokens across Base and Solana directly on Twitter by simply liking or replying to a tweet from oracle agent which mentions a specific ticker.

(Tokens are whitelisted when the agent tweets about them, and can be submitted for consideration via DM.)

Of course you need to set the agent up first which can be done on their website:

- You connect your Twitter
- Fund your wallet on Base or SOL
- Set the default buy amount, just like you would with Noice previously
- Start liking Oracle's tweets to buy tokens

There are two types of buys: regular spends, triggered by liking tweets, and super aligned buys, triggered by commenting "aligned" under an Oracle tweet.

The platform also has its own token, ORACLE, which is paired with NOICE. The Noice team has stated that the ORACLE token is simply meant to be a proving ground for the platform, rather than the central component of it. However, it does operate from a $20K treasury, charges a 1% swap fee on both likes and "aligned" buys, and uses all realized profits and fees to buy back and burn ORACLE.

While eliminating friction may be interesting to some, what has my attention here is how Oracle looks set up to be an emerging curation layer.

In its vision, Oracle details an upcoming scout program, where users can tag Oracle to identify new founders for Noice and earn from successful referrals. Selected founders, in turn, would be able to launch tokens directly through Oracle and embed buy actions into their own timelines, just like we can now do with Oracle.

Further, given the @jessepollak beta trial, I expect this could extend to not just founders, but also to CT personalities, though I could very much be wrong.

Closing Thoughts

While Oracle is a fun mechanism, I still hold skepticism, given that similar experiments, most notably Solana Blinks, tried to embed blockchain-native actions into Twitter and failed to take off.

Granted, Blinks did depend on stricter user requirements, including Phantom installation and specific feature toggles, which may have limited its reach. Oracle's simplicity could give it a different trajectory.

That said, beyond taking a shot on a bigger court, a defensible instinct, I'm still unsure why Noice prioritized Twitter over @farcaster_xyz, which feels structurally better suited to this behavior and is also where Noice has already shown traction.

Overall, Oracle reads as a proof of the (seemingly) most promising development trends in crypto right now: AI and internet-capital markets. It emerges from genuine momentum and demonstrates the expansion a team can pursue once they've developed a product actually in demand. I'm excited to see what will come.

[BeaconLayer Podcast @BeaconLayerCast]

Crypto’s social x money wave keeps heating up with content coins, creator tokens, mini-apps, and tipping.

The latest standout? Towns — a group chat app on Base where communities (free or paid) can earn, trade, and run bots right inside the convo.

If onchain group chats that actually make money sound like the future, this one’s worth watching👇

~~ Analysis by @kenzimori ~~

What Is Towns?

@townsapp is a messaging protocol and app designed around a straightforward but bold idea: Your group chats should be able to move money. Your wallet should be native to the chat. And creators should be paid directly by their communities.

Under the hood, Towns runs on a custom L2 chain for messaging, offchain stream nodes for real-time decentralized chats, smart contracts on @base that handle payments and memberships, and the $TOWNS token, which secures the Towns Network via staking.

As a user, all that complexity is abstracted away. What you see is Discord-like chats with built-in wallets, onchain paywall support, native tipping, and customizable bots.

Why does this approach matter? Crypto activity is already socially driven. Towns embraces this reality by making chats themselves into economic spaces. You don't leave a conversation to transact; you transact inside conversations.

Plus, it's yet another onchain business model for creators and communities to consider. Anyone can spin up a Towns chat, deploy their own bots, and kick off new revenue rails for their audience.

How to Try Towns

Another pro with Towns is that it's simple to dive into:

➢ Head to app (dot) towns (dot) com — Log in via Privy by spinning up an embedded wallet linked to your Google, Twitter, Farcaster, Rabby wallet, etc. If you want funds for joining chats, tipping, or trading, click the wallet icon in the top right to pull up "Deposit" and "Send" tabs for your embedded wallet.

➢ Explore Towns chats — The linked page is the app's main discovery hub; you can surf and join chats via recent activity, featured communities, top earning groups, and trending projects. Some Towns are free; others require a fee to subscribe.

➢ Create your own Town — In the app's left sidebar, press the "+" button to pull up the "New Town" deployer UI. Input your Town name, select from the "Free" or "Paid" options, and deploy. It's basically as easy as spinning up a new Discord server.

➢ Stake your TOWNS — If you decide to go deeper, head to the "Token" tab on the main Towns website and delegate your TOWNS to a node operator. This secures the Towns Network. Review operators' yield and commission stats, then make your pick and click "Stake."

The Big Picture

As social platforms have spent the past couple of decades becoming more extractive and closed down, Towns is moving in the opposite direction: open, onchain, programmable, community-centric.

Whether it can become the "Telegram of crypto" remains to be seen, but it certainly has potential, and if you're looking for a fresh corner of onchain social to explore, Towns is one experiment you can currently jump straight into. Explore a few chats and see what you think!

[BeaconLayer Podcast @BeaconLayerCast]

Takeover: The Onchain Fee Market You Can Fight Over

Takeover gamifies trading fees on Base through Harberger taxation—creating a market where 100 tiles representing 1% fee shares are perpetually for sale. Holders must pay 5% weekly taxes to maintain control while traders compete to snipe mispriced assets.

Here's how the mechanism works 👇

~~ Analysis by @kenzimori ~~

The Harberger Mechanism

@flaunchgg stands out for paying creator fees in ETH and tokenizing revenue streams as NFTs. Takeover builds on this infrastructure to create a PvP market for trading fees.

Every coin launched gets a 100-tile grid. Each tile represents a 1% claim on all trading fees paid in ETH. Own a tile to earn from every trade—until someone buys you out.

Harberger Tax. Tiles use Harberger taxation to ensure continuous circulation. Owners must set public prices, allowing anyone to buy instantly at that price with no negotiation.

Tax Structure. Holders post USDC deposits and pay a 5% weekly tax based on their listed price. These taxes fund $TAKEOVER buybacks through the Boardroom. Deposits must remain funded—run out and you forfeit the tile to the open market.

The Strategic Dimension

Success requires accurate pricing. Each tile has a fundamental value based on its parent coin's fee generation.

At the 5% weekly tax rate, a tile generating $10 in weekly fees has an equilibrium price around $200—where tax costs equal income. Price too high and carrying costs drain your deposit; too low and someone snipes your tile.

This equilibrium shifts constantly with trading volume. Dying coins become expensive to hold; runners attract bidding wars. Profitability depends on predicting volume trends and adjusting prices or exits accordingly.

How to Try for Yourself

Newcomers should buy into existing grids before launching new coins. The $FLNCHY grid (Flaunch's mascot) routes 80% of trading fees to tile holders, making it an ideal starting point.

> Browse. Find a listed tile on the 100-tile grid. The $FLNCHY floor currently sits at 68 USDC.

>Buy. Input your listing price and deposit duration. Total cost equals buyout price plus initial tax deposit. Confirm the transaction.

> Monitor. Earn 1% of trading fees in real-time ETH payouts. Adjust listing prices defensively as volume changes to prevent sniping.

Zooming Out

Takeover represents one of the first live tests of Harberger taxation with calculable yield—where mispricing delivers immediate financial consequences. AI agents are expected to join the competition soon, accelerating this economic experiment into a larger battlefield for automated strategies.