Every wallet and fintech is about to face the same question: Why don't we have self-custodial Lightning yet? Wavelength just collapsed the answer from "months of infra work" to a few API calls. No node. No channels. No liquidity management. Imagine Trust Wallet, MetaMask or Phantom adding instant bitcoin payments with the same effort as integrating any REST API. Imagine LATAM fintechs settling remittances in seconds for fractions of a cent. And here's the part most people will miss: the same API works for AI agents through MCP. Agents holding balances, paying per API call, settling with each other. No cards, no accounts, no humans in the loop. The agentic economy needed native money. It just got it. Self-custody included: users can unilaterally exit to the bitcoin blockchain anytime. Nobody's permission required. Bitcoin doesn't care if you're a human or an agent. Neither does Wavelength 〰️⚡️.

Yesterday, Cohort 12 went beyond the basics of Bitcoin and into the Lightning Network with @megasley 🇳🇬 — and he definitely brought his A-game. ⚡ The Dadas explored a key limitation of on-chain Bitcoin transactions: while the Bitcoin blockchain provides security and decentralization, putting every transaction directly on-chain can make frequent, small payments less practical. That’s where Lightning comes in. The session unpacked how Lightning enables faster, smaller Bitcoin payments and opened up practical use cases including remittances, merchant payments, payments for creators and freelancers, and micropayments. The Dadas also explored custodial vs. self-custodial wallets, communities already using Lightning, and ended with a practical challenge: make a Lightning payment this week. ⚡ Megasley took a technical subject and broke it down into concepts the Dadas could actually understand and engage with — taking them one step further into understanding the Bitcoin protocol beyond simply knowing the name. Bitcoin is the money. Lightning is one way to move it faster. 🧡⚡ #BitcoinDada #BitcoinAfrica #BitcoinEducation #BlockchainEducation

Stroke of a Pen I told Opus 5.5 to read my Bitcoin & monetary-history wikis & make a music video with code only. it used ElevenLabs for the track. Then a swarm of agents storyboarded and coded a 3:23 portrait reel. ~75 shots cut on the beats. Had to do 2 revisions - first the people looked like poorly animated stick figures & it rewrote the rigs. Then I said use matrix code to make it more interesting. Impressive!

NVIDIA's VP of Applied Deep Learning Research, @ctnzr, joins @matthew_d_white to discuss NVIDIA’s approach to open models, the strategy behind Nemotron, and the future of AI infrastructure ↓ Timestamps 0:00 - Why NVIDIA invests in open models 06:09 - Why small models need frontier models 09:32 - Hugging Face acquisition and NVIDIA’s open ecosystem strategy 13:53 - Why Nemotron goes beyond model weights 17:01 - Building a shared foundation for open models 25:35 - Why AI models need different licenses 30:23 - Why openness makes AI safer 36:32 - The limits of open model development

lightning⚡️

Grok Bot 🤝 Muse: “We won the mainstream AI agent race.” Open-source agents: not so fast. The next race is longer: real users, real tools, real payments. And agents need wallets: L402 + Bitcoin Lightning + stablecoins + Wavelength. The real race is just getting started. ⚡🤖

Just a small computer, at home, that becomes your cloud. Introducing umbrelOS 2.0. Out today.

built a malware scanner using laya – a tiny local decision model it can detect malicious files and directories before you open them

A lot of crypto use cases over the last decade were not things normal businesses actually needed. Payments are different. Cheaper, faster, 24/7 money movement is a real business problem. ⚡️

This is awesome !

Buenas! El 30 de septiembre tenemos nuestro bitdevs 15! Esta vez sera en @LaCryptaOk puedes registrarse en el link abajo y en la semana estare dejando los temas para que los lean 🙌

every time I hit approve on muse or instinct I feel the bitcoiner inside me slice off a piece of his soul so instead of sleeping tn I'll be wiring Wavelength + some more L402 endpoints across my agents to hopefully claw back a modicum of his respect

Just a reminder that GLM 5.3 Flash, DeepSeek V4.1 Flash, Qwen 3.8 Next Flash, and even Qwen 3.8 27B are all outperforming (in both intelligence and capabilities) every model that was considered "frontier intelligence" in Xmas 2025 (just 10 months ago) Opensource AI is on fire

IT security in 1990s

La “tasa de usura” en Colombia tiene buenas intenciones, pero termina perjudicando precisamente a quienes busca proteger. La explicación es contraintuitiva. Prestar tiene riesgo de impago. Si una persona tiene poco historial crediticio, es más difícil evaluar ese riesgo. Cuando la tasa máxima permitida no alcanza para cubrirlo, el banco simplemente niega el préstamo. El resultado para esa persona no es crédito más barato: es quedarse sin crédito. Y sin alternativas formales, termina en el “gota a gota”, pagando muchísimo más y exponiéndose a amenazas y violencia. El informe de ANIF y Colombia Fintech de 2025 cita que más del 65% de los adultos en Colombia no tenían crédito formal. Y estima tasas anuales promedio del “gota a gota” de alrededor del 382% para personas y del 666% para mipymes. Esa es la paradoja: un límite diseñado para proteger al consumidor termina cerrándole la puerta del crédito formal y lo deja expuesto a préstamos mucho más caros. ¿Cual es la solución? Eliminar el tope de intereses Y aumentar la competencia, exigir transparencia en los costos y mantener una protección firme contra los abusos. Quitar el tope, por sí solo, no basta. Gracias a desarrollos tecnológicos, Colombia tiene una gran oportunidad para impulsar esa competencia: • Bre-B puede facilitar que entidades digitales como Nu compitan con los grandes bancos sin necesitar una red de sucursales. Esto significa que pueden venir muchos mas competidores al sistema. Que gane el mejor! • Los pagos digitales dejan un historial que, con autorización del cliente, ayude a evaluar a personas y negocios que antes eran invisibles para el crédito formal. • El "Open Finance" permite que cada persona comparta su información con otras entidades para conseguir mejores ofertas. ¿Quién perdería con más alternativas formales? El “gota a gota”. La “tasa de usura” no ha funcionado en ningún país del mundo, y tampoco ha funcionado en otras industrias donde el gobierno decide por decreto “el precio justo”. Desde las políticas de alquiler de apartamentos en Nueva York o Buenos Aires, que dejaron miles de personas fuera de sus apartamentos y, al mismo tiempo, miles de apartamentos vacíos, hasta la decisión de Venezuela hace décadas de fijar los precios de los alimentos en los supermercados, que dejó los estantes vacíos. A pesar de ser contraintuitivo, el libre mercado es muchísimo mejor para encontrar el precio adecuado de un producto que una política de gobierno.

“Si el Estado no somos todos ¿A quién le estamos pagando impuestos? Si el Estado no somos todos ¿Unos viven del esfuerzo propio y otros del esfuerzo ajeno? Si el Estado no somos todos ¿Mañana los impuestos serán un robo?” @juanmgiraldor

“Models have been superhuman at chat for years. So where is all the automation?” I took @CompleteSkeptic’s question literally and built a small experiment. Jev vs. six LLMs playing Snake. Same seed, same four-way decision space, 60 seconds each. Jev: 21 pts / 204 decisions Haiku: 9 / 73 Luna: 5 / 43 Sol: 5 / 40 Kimi K2.7: 5 / 34 Opus: 3 / 25 Kimi K3: 2 / 14 My first thought was: maybe Jev is simply buying speed with worse decisions. But at equal move counts: 25: Opus 3, Jev 3 40: Sol 5, Jev 5 43: Luna 5, Jev 5 73: Haiku 9, Jev 10 Then I noticed something I think is more interesting. Jev’s median decision took 263 ms. Its 16 least-confident decisions took 280 ms. Uncertainty did not really turn into more computation. It showed up as uncertainty. This made me wonder if we sometimes conflate intelligence with deliberation. A lot of software does not need a long answer. It needs a small decision, over and over again — and some indication of when that decision deserves more thought. There is potentially a very large design space here: payments, fraud, routing, logistics, agent loops — anywhere decisions arrive much faster than humans could ever inspect them. Perhaps the path from chat intelligence to large-scale automation is not making every decision think harder. Perhaps it is learning where thinking is actually worth spending. github.com/angelgalvisc/snak… Really interesting work by @typesafeai and @CompleteSkeptic. Lots to explore here.

ppl like to talk about "taste", but IMO what's important is vision if you can sense/see/feel the _shape_ of a solution or problem, then in our new experimental era of software, once you can see the Opening Thread, you're nearly done by consuming more high-signal non-slop fundamental papers/books/talks, you train your Wetwear pattern matching system (satori!) to recognize more and more shapes, to further identify new Opening Threads, leaving solutions you never actually had the time to work out in your wake

Automated payments are going to put more pressure on payment infrastructure than people think. It is one thing to support occasional human payments. It is another thing to support constant, small payments where cost and reliability matter every time. That is a place where Lightning starts to make even more sense.

Grok bot using self-custodial Lightning ⚡️! @grok @bot