Mesh and @Visa are partnering to make crypto and stablecoin balances spendable anywhere Visa is accepted: 175M+ merchant locations, straight from the wallet apps people already use. No separate card. No cash-out. Nothing for merchants to change. na2.hubs.ly/H08sd-G0

Incredibly big milestone for @bamazizimesh and @meshpay, expanding stablecoin payments to merchants globally direct from wallet with @Visa. Love to see them continue to crush it.

Opening night at @money2020: our CEO @bamazizimesh sat down with @alexblania, fresh off the launch of @world_id for banking and finance. When agents start paying on our behalf, proving there's a real human behind them becomes core payments infrastructure.

Jeremiah Smith is the best WR in CFB history.

“Banks could engage in pro consumer behavior to stay competitive”

RIP the agent bank run slop Sep ‘26 - Oct ‘26

Santi might be a little more bothered about AI stocks this week than he was last week

.@HadickM on @MTSlive re: agents + money: “When you’re built on top of Plaid, you have a bunch of failures, failure rates at 2% or so for ACH. And there’s a lot of capital inefficiency in it as well.” “If you ask your LLM what bank should you use that is agent friendly, they’ll tell you Mercury. They won’t tell you JPMorgan.” “A financial institution that controls those different accounts and is able to monetize those different accounts, that is AI first and agent first, that is where you’re likely to see a lot of this happen.” “The concern is the surplus will continue to shift from banks to fintechs.”

I've been looking for an excuse to work with @HadickM and @Nick_van_Eck, two of the smartest people in stablecoins. They bring together a great investor + founder perspective on the space. Excited to share that they'll be joining us as our hosts for @MoneyCodePod Season 2 🙌 Now, who should we interview? Drop us a line at @thestablecon.

New look, new hosts, new season... and now with the absolute best interviews of those entrepreneurs and executives building the future of finance and capital markets.

Join us this Sunday in Vegas for Stablecoin Day. What does it take to put stablecoins to work in everyday business? We’re talking banking, global payments, private wallets, local payouts and compliance. Hear from @KevLeht, @pankajbengani, @SamiStart, @HadickM, Derek Woods and Caitlin Barnett, with more speakers to be announced. Plenty of time to compare notes, meet potential partners and connect with the TWIF community. 📅 Sunday, October 18 | 12-5 PM 📍 Mercato Della Pescheria, Las Vegas Request your complimentary pass: luma.com/1yhb99bu Powered by @alcove_wallet, @chainalysis, @realMeld and @withpersona.

Feels like there is a new Mesh product on average every 36 hours. Insane shipping speed.

Ironically, this poll is a prime example of how polling is such a significantly worse form of information gathering than prediction markets. They specifically asked respondents if they opposed “betting on election results”, which obviously is actively asking the respondent to have a point of view on gambling specifically and then using that to extrapolate to broader PM use cases. There is endless research about the use of negative language and how that affects polling. If it asked, “do you oppose being able to hedge political risk”, this would swing significantly the other direction. A use case we are seeing a ton of institutional demand for and is likely to be far bigger the retail use case (which is also not necessarily gambling!) in this election and the next. You’d think with how thoroughly embarrassed the pollsters have been in recent years that they would have learned their lesson by now.

Variational is now the largest on-chain venue by volume for RWA derivatives (perps and swaps). We're just getting started.

These numbers are eye popping and I would bet there are significantly worse unit economics among a bunch of the well known highly funded AI startups.

I posted this 4 days ago. We're at 500B tokens/day now.

Stablecoins need to work in all the ways people use money today. Rain wallets let users get paid, earn, spend, and send money. All from one balance. Available through a single integration.

Mesh has acquired @Bluvo_co, a developer of exchange connectivity technology. Same mission, now at network scale. Welcome to the team, Luca Dalla Chiara and Florian Sabani. na2.hubs.ly/H08l41m0

Muse, Grok, Dots, and Instinct agents can now use stablecoin cards. Rain's scoped cards on @oobit give agents a single-use card for each approved purchase, with spending limits and optional merchant restrictions. Agentic commerce runs on Rain.

I should probably also mention, @dragonfly is investing in @typesafeai's latest round! :) My quick thesis on decision models, since I've been having a lot of conversations about this lately: LLMs are amazing because they are fully general. You can use them for pretty much anything. But they are slow, expensive, and too unreliable to embed in software at runtime. Unless your product is literally chat, real-time LLM calls are unworkable for the majority of use cases. (Especially in finance!) Decision models change that. To abuse an analogy, they are the ASICs of LLMs. They give up the generality of an LLM, but in exchange for constraining your outputs, you get MASSIVE gains in cost-efficiency, throughput, and latency. In the early days of the AI boom, people understandably tried to shove LLMs into everything. But like with ASICs, you will eventually constrain your problem enough to burn it into the machine. Once your problem is an if-else, a buy or sell, a label, a choice, or a well-defined state machine like an onchain smart contract, a decision model will clobber the LLM on cost, speed, and reliability. Most systems are not language-in, language-out. Going forward, LLMs will write our code (since coding is high-dimensional and general), but that code will embed decision models to make runtime software intelligent and capable. That's the bet we're making with Typesafe. They identified this path and proved that it works. 25%+ of the Fortune 500 is already using Jev, and it launched less than a month ago. Of course, there are murmurings on the timeline that decision models are easy to copy. OpenAI, Microsoft, Amazon, and Cloudflare have all launched decision models in the weeks since Jev launched, realizing that this is going to be an enormous market and they don't want to lose it. To be sure, anyone can lop off the last layer of Qwen, fine-tune it to create a classifier, and copy the Typesafe SDK's API. But there's a lot more that goes into a world-class decision model, and there's a reason why Jev is still cheaper, faster, and more general than these other alternatives. (Decision model benchmarking started all of 10 minutes ago, but if you ask devs who've tried them, they'll tell you Jev is far and away the best right now.) There's no guarantee that lead lasts forever though. Decision models are too big of a market to be able to coast into a monopoly. From here, the onus is on them. They have to keep training better models ($$ helps), ship great products, serve their models better, faster, and cheaper than anyone else (many more tricks there), and win developers with their community-first philosophy. But they've been at this for years and have a lot up their sleeves (they were expecting a bevy of copycats if Jev took off). Diogo knows from his years at OpenAI how to sustain a research advantage. And as with all startups, focus is how you beat the incumbents. Super bullish on this team. And if you're a dev and you haven't tried Jev yet, get on it.