Kobeni.hi (@HorusVC) header
Kobeni.hi (@HorusVC) profile picture

🚀$hyper

Texas
Joined August 2023
Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)Photo by Kobeni.hi (@HorusVC)
Kobeni.hi retweeted
Eris
4
344
1
4,872
27,199
🚨 NVIDIA colapsa -10% en un solo día.
4
1
24
724
Kobeni.hi retweeted
If you use a crypto app in Mexico, starting Nov 30 the platform has to record who is on each side of your transactions and keep that data for 10 years. Acuerdo 115/2026 from Mexico's finance ministry (SHCP) covers the sender, the receiver and the beneficial owner when there is one: name, country and wallet or account. It also covers the date and time, the asset, the amount in crypto and in pesos, the operation type and the fee. The UIF, Mexico's financial intelligence unit, and SAT, the tax authority, can request it. The rule adds record keeping and reporting duties for platforms, with no new license or tax attached. For people sending remittances or moving USDC or USDT, it means a longer paper trail. You may have seen me post something similar in Colombia, where crypto platforms now have to report users' 2026 transactions to DIAN, the tax authority, with the first report due in May 2027. Both rules come from global standards. Mexico's follows the FATF, the international anti-money-laundering body, whose "travel rule" asks platforms to keep sender and receiver details with each transfer. Colombia's follows the OECD's crypto reporting framework, which lets tax authorities share that data with each other every year. If you send or receive crypto in Mexico, has your app asked you for more information yet?
47
61
23
367
41,439
Kobeni.hi retweeted
Well, I didn't expect putting Sunflower Land in D Tier to cause this much chaos lol I've received quite a few comments from Sunflower fans, with some even calling my tier list rage bait. So I guess I'll explain it one more time. Guys, seriously, learn to read the actual post before jumping into the comments. I literally explained all of this in the original post. Are we really that lazy now? lol First of all, this was NEVER a ranking of the BEST web3 games. It was my personal take on the most promising projects right now, based on my own interest, expectations for the future, and the overall sentiment I've been seeing across CT. Sunflower Land has been around for years, built a loyal community, and clearly still has a lot of passionate supporters. I respect that. But being a good, established game doesn't automatically mean it has to be S Tier on everyone's list. D Tier doesn't mean the project is dead, bad, or has no future. It simply reflects where I personally place it compared to the others right now. And actually, seeing how strongly the Sunflower community reacted was pretty interesting. Maybe I underestimated just how active you guys still are. At the end of the day, different opinions are what make these discussions interesting. No rage bait, just my perspective. And I'm always open to changing my mind when projects give me a reason to.
A web3 gaming tier list so you never have to think web3 gaming is dead Here are 22 projects I believe still have a lot of potential. A few things to clarify: I ranked Moku, OCH, and Axie Infinity as entire gaming ecosystems rather than individual games. They're among the biggest game ecosystems in the Web3 gaming space, and I expect a lot more to be built around them. This tier list is based on my personal interest in these projects, their future potential, and the community sentiment I've been noticing across X. A lower tier doesn't mean a bad project. It's simply how I see things right now, and that could change. There's still plenty to look forward to in web3 gaming. What other projects would you add to this list?
11
2
16
1,609
Kobeni.hi retweeted
we need more L1s! sui $390M raised - $33 revenue 24h berachain $142M - $27 sonic $100M - $20 movement $41.4M - $0.01 hedera $100M - $0 provenance $45M - $21 tempo $500M - $27 zcash $26M - $0 sei $35M - $0 unichain $178M - $0 ronin $160M - (-$9.32) taiko $37M - (-$82) manta $31.6M - (-$12) peaq $40M - $0 celestia $101.5M - $0 stacks $108M - $0 strato Chain $41M - $0 mythos $300M - $0.22 cronos $100M - $0 babylon Genesis $103M - $0 chia $69.4M - $4.45 $2.65B raised in total but combined 24h revenue just $29.36 guess which one shutting down next?
55
4
1
143
10,054
Kobeni.hi retweeted
🧬 THE FUSION CHAMBER IS LIVE! 🐜 Fuse 1–5 ants for a chance to get a rarer one! ✅ Success → Full lifespan! ❌ Failure → Ants are lost! 🎰 Feeling lucky? Go Double or Nothing! 👑 4 NEW MYTHIC ANTS ⚡ Lightning | 🍯 Jelly | 🧟 Zombie | 🪖 Military 🔥 Fusion-exclusive. Stronger than Legendary ants! 🎮 antropolis.fun/play 💊 pump.fun/coin/2XjNoVMEhKYzen…
36
274
3
351
2,433
🔥 Trending Solana P2E games — 24h market cap movers: 1. PlanetForge @planet_forgecom solana:CSE6QUG33774zUESuj9ox8Kdnd5yFCVkagCmokRupump (+13.3%) 2. AFK Heroes @AFKHeroesXYZ $AFKHERO (+12.2%) 3. Gable Guardians @Gableguardians solana:9uTekAxmx4rqBcR3iGkKmaBwVWMmi7TXEU41frBdBAGS (+10.8%) 4. Baked Beans V5 Classic Solana Edition @AGSCryptoGaming $AGS (+6.6%) 5. Rise of the Gorecats @GORECATS solana:UbESBaztbkxJRWxPcfDeK8Fft15igTbrv3sed1bsegM (+5.2%) solanap2e.games #Solana #GameFi #P2E
1
2
5
525
Kobeni.hi retweeted
Survivors @PlayCalamity – Mage Build Tier List I've put together my personal ranking of the mage build based on my experience and understanding of the game. I'll explain why I placed each build in its respective tier. Keep in mind that these are my personal rankings, and your experience may differ. Let's get into it.
3
2
13
514
Kobeni.hi retweeted
有位黄仁勋的疯狂粉丝收藏了1995-2025 三十年全部盘点 也算是时代的见证了
279
1,442
119
18,007
1,110,419
Kobeni.hi retweeted
Tips for Beginners in @PlayCalamity Survivors №1 Skills One of the most important strategies during your run is choosing the right skills I focused on upgrading all the skills from the Spellbinder talent tree So this tier list is for those who have chosen the same path • S-Tier: The foundation of your build - without upgrading these skills, surviving in the dungeon is nearly impossible (Arcane Explosion + Blinkcast + Mana Shield) • A-Tier: Since Fire Ball is your starting spell, upgrade it when no S-Tier skills are available. Arcane Wisdom can replace Blinkcast if it doesn't show up for a long time • B-D Tier: I almost never use these skills, as they waste upgrade opportunities that could go toward higher-tier skills Don't forget that you have 3 spell rerolls and 2 removals per run It's best to use them when the spells you need aren't available or when you get a Common rarity option
4
3
21
1,088
🚨21 yaşında bir öğrenci 50 $ harcayarak 1.000.000 $ kazandı Okul masraflarını ödeyemediği için, boş bir site açarak piksel satmaya başlamış • Siteye 1 milyon piksel koydu • Her piksel $1 İlk 4.700 $ çevresinden geldi, 400 $’ı reklam için harcadı ve BBC haberiyle site patladı • 2 haftada 250.000 $ gelir • Zirvede günde 100.000 $+ gelir • Son 1.000 piksel eBay’de 38.100 $’a satılıyor Para kazanmak için dahi olmaya gerek yok. Sadece harekete geçmek lazım✍🏻
3
3
49
48,328
Kobeni.hi retweeted
Finally caught the Micropinna Chapter Marvel Fish! 🎣🐟 Another exciting milestone in Chapter 16: Cave Dwellers! This beauty gives +0.1 Obsidian per harvest. 💎 Since it’s Spring, I switched to Fishing Lure/Red Wiggler + Cabbage to target the Coelacanth! One more rare catch secured. Let’s go! 🔥 #SunflowerLand #CaveDwellers #SunflowerLandFarm #SunflowerLandWeekly
🤖 Made with AI
16
2
45
1,664
Kobeni.hi retweeted
Pigs aren't profitable for early- to mid-game players either, at least not without the right skills or NFTs. I know, I know... Deox never learns his lesson, does he? 😂 But I had to satisfy my curiosity, even if it means slowing down my progress and my rush to Vulkano. The first few times I fed the pigs, it was actually profitable. But at level 3, they started asking for 4 Basic Feed, which already costs more than the truffle itself. With the Mug, I managed to reduce the feeding requirement for one of them from 5 to 4 lol. It also says it helps prevent them from getting sick. Still, I'll keep raising them slowly. I noticed that sacrificing one pig can give you 1–12 truffles. Is that right? Can I actually get 12 with some luck, or am I misunderstanding how it works? I also saw a post from @ChuckFresco showing that high-level pigs can generate a pretty good amount of coins. It would be interesting to level them up and see for myself. Maybe around level 13. 🐷 @0xSunflowerLand #SunflowerLandWeekly
7
1
10
1,038
Kobeni.hi retweeted
11
1,270
21
21,803
199,803
$55k ➔ $3.25M a day $18k ➔ $3.8M a week Two web3 games Same trick Weeks apart 👇 Here how these games made players money
2
4
10
2,634
1:When I made this call, $RON was at 0.30. Just as predicted, it bounced right off 0.035. WTF is this Binance FUD People don’t realize $BTC and $ETH are purging the market.These dumps are executing 1000 of useless alts. This isn't a black swan it's a cleanse for next Bullrun.
8
8
43
3,959
NEW: HFT market maker @0xLoris criticizes Hyperliquid's Dutch auction read priority system, arguing it gives firms like @wintermute_t an unfair execution advantage. According to Loris, the mechanism allows priority buyers to see fills 50–60ms earlier and hedge on Binance or Bybit before competitors. He argues this could concentrate altcoin market making among the deepest-pocketed firms, reducing competition and hurting long-term liquidity. Loris calls for a more level playing field that allows more market makers to compete on Hyperliquid.
btw i think you may be arguing against a chatgpt X bot but for the sake of continuing the conversation... > Wintermute is and has always been a net positive for crypto i broadly agree with this. wintermute is synonymous with functioning crypto markets. 'dont hate the player, hate the game'. if anyone blames them for their bags going down, maybe they should ask if their bags ever had intrinsic value in the first place. > First of all it’s a read priority, which means that they pay for the access of incoming tx. They don’t get the ability to execute earlier than others ok. so, the crux of the issue is not about picking others off on HL. the problem is this: they get to see when their orders will be filled, first - always, 50ms (60ms now) ahead of anyone else on hyperliquid, and then HEDGE before anyone else on other exchanges. in doing so, they can size up to the point where they squeeze every last drop of juice out of smaller markets on HL, by consuming ALL of the profitable hedging liquidity on binance (et al). the only reason they WOULDN'T do this is if they were being charitable and letting others take some pnl (or maxed out on position size). and wintermute is not a charity. if there's only $100,000 of profitable hedging liquidity for UNI perps out to 50bps on binance+bybit+okx, then they can just quote that full amount / depth knowing they can consume ALL of the hedge liquidity, 100% of the time (simplifying a bit) this means alts converge to having 1 market maker. which is structurally - and this is not debatable - bad. they could even quote wider than other MMs and still hedge ahead of them because they get to see their fills that much earlier. 60ms is an eternity in HFT. round trips between tokyo boxes is under 1ms, so that's 60+ round trips worth of network time. while everyone is listening to a dial tone. so the actual question is, are we ok with having only a single market maker quoting markets on hyperliquid? i would strongly argue no, we are not. it's not democratic, it's not sound microstructure. it's bad for hyperliquid, and it's bad for crypto. > The 227M PnL is irrelevant because it was generated by a ton of alpha instead of just them having most of the priority slots agree. also, for the sake of argument, they might have lost 228M on their hedges against these positions. they didn't, and the original number is probably not that far off, but you can't make any profitability claims based on their account history on a single venue. also wintermute is (mostly) not wearing naked HL positions w/o hedges. > Why didn’t you mention on your other tweet that all crypto got nuked? You blame WM and HL for the 82% liq collapse but you don’t see you mentioning that nance is also down like 60%+ on that same period just like every other exchange lol that's still a big difference, but causality here is difficult to prove. factually, i can tell you we can see binance's hedges hitting the binance tapes 60ms+ ahead of ours, and size-match it to their HL wallets fills. this is what is driving smaller MMs out. we have recently cut our quoted symbols in half on HL from 120 to 60. > It literally converts MMs trading edge in fees no, that's not how it works. as mentioned, they could even quote *wider* than they otherwise would, because they have guaranteed hedge liquidity. also, it's similar to how 0 fees doesn't mean 0 spreads. market makers do not just eat fees, they widen spreads according to costs. there are shades of gray here, it's not binary, and i am open to a multiple equal-speed slot architecture with open, easy peering access for any slot winning wallet. that was my first recommended redesign in fact. but that may not be possible given the network/architecture. > Abolishing the auction doesn’t change anything in fact it might actually make it worse. Do you know how much funds spend on infra every year? Take it away and now have the same exact thing happening but behind closed doors and with fewer participants no. our all-in HL-related infra bill was around $15,000 last month, and i know for a fact we are within 1ms of wintermute's HL latency, absent read-priority. wintermute spent $828,676.86 on read priority in the past 30d. this fact is available from hl core chain data. $828,676.86 > $15,000. could argue the point on engineering, but there's not that much engineering required once everything is in place. furthermore i doubt they reduced their infra costs with read priority, but you'd have to ask them > The auction is better for the smaller MMs because they can’t afford the infra cost to race for ms, but they can afford to buy temporary priority no. this was the stated intention, but i struggle to see how they didn't see the obvious outcome of the deepest pocketed player dominating the auction. which makes me think this was intended as a v0 that just hasn't been updated yet. so now it's time to update or simplify (abolish). > I never personally worked in HFT/MM but Id argue that since it massively helps against adverse selection it also allows MMs to quote with much deeper liq at top of book and sometimes maybe even tighter spreads the incentive could also be the opposite, as stated above. my observed behavior, as an HFT MM competing directly with them, staring at this stuff 90 hours per week, is that alt markets are and will continue to converge to having a single market maker because nobody else can quote profitably. if wintermute blows up, or decides to stop quoting, or their systems take a dump for 12 hours on a wednesday, those markets go quiet. bad. > 1. WM is buying 76% of slots because they’re better than everyone else > 2. They’re buying it so others don’t have access to it > Sorry to say but in both cases it’s based af lmao i more or less agree. they're just playing the game, and winning. i would again like to state unequivocally, wintermute is NOT the problem here! they're just playing the game and winning. but if we're going to house all of finance the game cannot be designed in this way. it has serious risks, and it is not good enough. i've spoken to many smart people about this and they all agree. respectfully, the only one who disagrees is the one who's winning the game. hyperliquid should stick to its day-1 values and be a democratic, open ecosystem that is NOT designed for the few. it should be designed to support the deepest liquidity, with the most participants, to serve a global market. and under the current design, it's simply not so.
2
2
1
35
4,585
Stone is flying 😄 Cave impact I believe @0xSunflowerLand
5
3
22
1,152
Kobeni.hi retweeted
Çok güzell ağlicam
İnstada bir kız kendine kelebek modelinde hırka örmüş o kadar güzeldi kii😭
502
11,780
1,152
145,840
2,703,321
Que impresionante es @bot Me creé una flota dedicada a fortalecer mis sistemas de Market Making y le marqué 17 cuentas Chinas que sigo en X (ahí está toda la data de Trading algorítmico) Me resumió todos sus posts de MM y HFT en un PDF de 107 hojas 🔥 🔥 Que belleza
🔥 Guía completa de cómo usar GrokBot desde 0 para que cualquier persona pueda empezar GUARDÁ este post desde ya porque va a estar cargadísimo de valor, luego de investigar qué es lo que mejor funciona probándolo exhaustivamente a GrokBot Cómo les comenté, me pasé de Claude a Grok al 100% El Viernes les mostré el desk de inversión con 10 agentes en Grok Bot y hoy les dejo el post único: cómo usarlo si nunca tocaste un agente, y cómo migrar si ya venías de Hermes o OpenClaw (sistemas de agentes que instalás y mantenés vos en tu propia computadora) Lo que a mí me llevó uno o dos meses con Hermes, @bot me lo armó en horas. Si estás empezando te cambia la forma de trabajar, y si ya te das maña es más simple y más rápido para el 99% de lo que ya venías haciendo El error de casi todos, y por eso les sigue pareciendo un ChatGPT con pasos extra, es tirarle tareas sueltas. Un bot con lista de pendientes no escala, un bot con un puesto sí, y esa diferencia es la que separa a los que lo usan de los que lo entienden El sistema entero, el que está usando hasta gente de SpaceXAI, cabe en una sola línea Rol -> Tools -> Skill -> Routine -> Team (un puesto, las herramientas de ese puesto, cómo se hace, cuándo se corre solo, y el equipo) Abajo va el método y al final todos los videos y posts que más me sirvieron para setearlo Qué es, sin humo Grok Bot no es otro ChatGPT. Es un equipo de agentes que vive en una computadora en la nube (los servidores de SpaceXAI), con navegador, archivos y terminal (la línea de comandos). Se entra a tus aplicaciones y sigue trabajando aunque vos cierres tu computadora La diferencia con Hermes y OpenClaw es bastante concreta. En Hermes y OpenClaw vos ponés la máquina (una computadora, un servidor o una Mac Mini encendida todo el día) y la memoria y los instructivos son archivos que podés abrir y editar a mano. En Grok Bot la máquina te la dan: hay un solo computer por cuenta (una computadora en la nube compartida por todos tus bots), así que tenés menos control y mucha menos fricción Si querés personalizarlo al 100% Hermes gana. Si querés que el 99% del trabajo salga ya, Grok Bot Y el talento no es escribir mejor el pedido. El talento es armar el equipo: quién hace qué, quién le reporta a quién, y dónde te frenás vos a dar el OK Si estás empezando: de 0 a un equipo en 10 minutos Qué necesitás - SuperGrok Plus o Heavy, o Cursor Pro+ o Ultra (los planes que hoy incluyen Grok Bot) - La aplicación de escritorio en Mac o Windows, y el celular para aprobar desde afuera - Arranque: x.ai/bot - Descarga: cursor.com/bot/onboarding Los 5 pasos van en este orden y no conviene saltear ninguno 1) ROL New, o Cmd/Ctrl+N (el atajo para chat nuevo) -> Create new agent -> Edit Profile Nombre, un solo trabajo, y una descripción de cómo tiene que trabajar. “Asistente personal” no es un puesto, “Analista de noticias del S&P 500” sí. Un bot es un trabajo y un resultado del que es dueño, no una caja donde tiras todo. Esa es la regla de oro 2) TOOLS (herramientas) Settings -> Plugins (conectores oficiales a tus aplicaciones: Gmail, Calendar, Discord, GitHub, X) Conectá solo lo que ese bot necesita. Primero el conector oficial y después el navegador, porque el conector es más estable que ponerlo a hacer clics 3) SKILL Instructivo reutilizable: los pasos, las reglas y el formato de una tarea, para no explicarla cada vez) Hacé el trabajo una vez, corregilo, y cuando sale bien pedile: “Guardá este proceso como skill. Incluí fuentes, formato y qué requiere mi OK” 4) ROUTINE Automatización: le decís cuándo correr ese instructivo, por horario o por un evento Recién ahora aparece el cuándo. Un ejemplo: “Todos los días a las 8:00 AM, corré el skill Daily mercado. Publicá el watchlist en esta conversación. Si no hay datos, avisá. No uses información vieja”. Test run siempre (una prueba real: el bot hace el trabajo de verdad, no es un simulacro) 5) TEAM (el equipo) Un CEO (el gerente: vos le hablás solo a él y él reparte el trabajo) y especialistas abajo. El prompt que a mí me desbloqueó todo, pegáselo al primero “Esto soy yo y esto quiero lograr: [tu vida y tu trabajo en crudo, una o dos páginas]. Vos sos el CEO. Armá el resto de los bots. Yo solo te hablo a vos. Nunca envíes mails, operes, publiques ni gastes dinero sin mi OK. Cuando un trabajo sea distinto, creá un bot especialista y asignáselo” Después, reverse prompt (darle vuelta la tarea: que él te escriba sus propias reglas) “Escribime tu system prompt: quién sos, cómo trabajás, qué no hacés, cómo me reportás. Guardalo en tu descripción” Eso es el brain dump (tirarle todo el contexto de una, y que él se arme encima). Es el arranque más rápido que vi En el compositor (donde escribís) usás / para skills y @ para bots, grupos, routines y plugins Skills, routines y Teach a task Hay 3 bloques y no conviene mezclarlos - Skill = CÓMO se hace. Pasos, reglas, formato, qué requiere tu OK - Routine = CUÁNDO se corre - Teach a task = le mostrás, no le explicás. Abrí la computer view (la pantalla de su computadora en la nube) -> Teach a task -> hacé el flujo una vez, hasta 10 minutos -> revisá el skill que arma -> testealo antes de agendarlo El orden correcto es tarea suelta -> corregir -> skill -> routine. Nunca automatizar el primer resultado El brief de 5 puntos Esto es lo que separa un bot útil de uno que inventa. Cada vez que le pidas algo, cinco cosas - Outcome: qué tiene que quedar terminado - Sources: de dónde saca la información - Constraints: qué no puede hacer - Deliverable: qué te tiene que devolver - Stop: dónde se frena para que vos revises La primera prueba, sin entrar a ninguna cuenta: adjuntá un PDF y pedile un resumen. Si no respeta el alcance, no cita evidencia o no se frena, no le des tus aplicaciones todavía La regla de las 4 corridas Las corridas 1, 2 y 3 van a mano, y cada una tiene que dejar una instrucción nueva: qué recortar, cómo ordenar, dónde frenar. Si el primer reporte es enorme, no empieces de cero, decile qué dejar, qué sacar y qué acortar. Recién la cuarta se agenda como routine Y el resultado tiene que caer donde vos ya mirás. Un reporte automático en un documento que nunca abrís no sirve. Yo lo mando a Discord El ciclo de trabajo, no el de un chat señal -> juntar contexto -> ordenar -> borrador -> tu OK -> ejecutar -> aprender Si ya venías de Hermes u OpenClaw No reescribas todo. Pedile a Grok Bot que hable con tu agente de Hermes, copie el flujo de trabajo y lo adapte al sistema de @bot. La traducción mental es esta - Profiles o agents aislados -> Bots (hasta 50 bots y grupos en total) - MEMORY.md o SOUL.md (archivos de memoria y personalidad) -> la descripción y la memoria del bot, acá no es un archivo que puedas editar a mano - Skills en Markdown -> Skills + Plugins (Settings -> Plugins) - Cron o heartbeats (tareas programadas en tu servidor) -> Routines, hasta 50 por bot - Una máquina por agente -> un computer compartido por la cuenta - Cualquier modelo -> corre en Grok - Vos mantenés el servidor -> cero mantenimiento Lo que vas a extrañar: aislamiento entre agentes, editar la memoria a mano, y mezclar modelos (uno caro para pensar, uno barato para ejecutar) Lo que ganás: chats de grupo, pasar trabajo sin volver a iniciar sesión, Teach a task, y que el equipo trabaje con tu computadora cerrada El computer, esto es lo que nadie explica Todos tus bots comparten el mismo navegador (cookies e inicios de sesión), los mismos archivos (/workspace, la carpeta permanente de esa computadora en la nube) y las mismas credenciales de terminal. Por eso pasar trabajo es gratis, y por eso un bot mal definido también puede tocar lo que no debe Reglas que yo uso - Nunca pegues una contraseña ni un código de doble factor (el de 6 dígitos del celular) en el chat. Si pide entrar a una cuenta, tomá el control de Agent Computer (la pantalla de su computadora), lo hacés vos, y le devolvés el control - Plugins primero, navegador después. El conector es más estable que los clics - Archivos importantes siempre en /workspace, no en carpetas temporales - Las pantallas de cada bot no son una pared de seguridad. Si un bot no debería ver un inicio de sesión, no lo cargues en esa cuenta Modo pro: flota chica Empezá con el equipo más chico que sirva - Un CEO, el único con el que hablás - Especialistas solo cuando el rol es estable - Group chat (un chat de grupo entre bots) cuando el traspaso tiene que verse - Tu OK para cualquier acción externa: mail, operación, publicación, borrado, pago, producción En el desk de inversión yo armé 10. Uno mira noticias y manda el daily. Cada 30 minutos la flota chequea el mercado y tira watchlist al Discord. Paper en Alpaca (cuenta de práctica, sin dinero real). Todavía no live Cómo lo tengo pensado, para que se entienda el organigrama sin copiarle el de nadie - Yo arriba - Un CEO, el único con el que hablo - Desk de mercado: noticias del día, watchlist (lista de seguimiento), riesgo, y ejecución en paper - Operaciones: reportero a Discord e investigación Y tres reglas que me están salvando el equipo. Un bot que decide si hace falta un bot nuevo, un skill o una tarea suelta, porque si no terminás con veinte clones. Los agentes se verifican entre sí: si el de noticias marca una señal, el de riesgo la cruza antes de que ejecución mueva un dedo. Y shadow mode (modo sombra: el bot propone y no ejecuta) hasta la segunda prueba buena El lunes, si hay amor, hago el paso a paso de cómo armar un desk de inversión parecido a lo que hice Tokens, límites y no romper nada Grok Bot es una aspiradora de tokens (los créditos de uso de inteligencia artificial). El límite semanal se quema rápido si lo ponés a “aprender” las veinticuatro horas. Enseñar es caro, las routines son baratas, así que enseñá uno o dos días y después agenda. Grok, Cursor y Grok Bot tienen créditos separados: en el chat y en Cursor vas holgado, en Bot no Cómo no quedarte sin límite a mitad de semana - Test run SIEMPRE antes de habilitar una routine. El test hace trabajo real - Borrador primero, ejecución después. Mails, operaciones, tweets, borrados y pagos = tu OK - Si una fuente no está, que reporte el fallo. Nunca que invente con información vieja - Si te ausentás varios días, revisá que no te haya pausado las routines Un bot nuevo necesita contexto y acompañamiento en las primeras corridas. Eso no es un fallo del sistema Errores que estoy viendo en todos lados - Un bot que haga de todo - Tareas sueltas en vez de un puesto - Automatizar el día uno. La cuarta corrida, no la primera - Quince chats en paralelo en vez de un CEO - Pegar secretos en el chat - Dejar el resultado en un lugar que nunca mirás - Escuchar cada mensaje de Slack, eso se come el límite - Tratarlo como Hermes: acá no vas a buscar un MEMORY.md con el buscador - Usarlo de único revisor en código donde equivocarte sale caro. Para código pesado yo sigo mezclando Grok 4.6 Hyperfast y Fable 5 en Cursor, Claude sigue siendo el mejor para revisar proyectos grandes - Querer una agencia de varios clientes en una sola cuenta. Hoy el entorno es compartido, bien para uno, mal para cinco clientes con inicios de sesión distintos Lo que más me sirvió El sistema Rol -> Tools -> Skill -> Routine -> Team, el más claro en 60 segundos: teapawt.org/ScottyBeamIO/status/20… Cómo hacerlo bien: brief de 5 puntos, 3 corridas a mano, la cuarta se agenda. 21 bots para copiar: teapawt.org/heyrobinai/status/2090… View article El talento es armar el equipo, no el pedido: teapawt.org/rewind02/status/209086… Los 10 bots más útiles para instalar (Bot Advisor, CEO, Automation Scout y más): teapawt.org/aiedge_/status/2090610… Guía de 15 minutos de Miles Deutscher: teapawt.org/milesdeutscher/status/… View article Setup de Alex Finn: youtube.com/watch?v=vrgO4D_m… Primer contacto: youtube.com/watch?v=lc6hKU4B… GrokBot es el mejor para cualquier persona y te explica a edtalle la razón: youtube.com/watch?v=vrgO4D_m… 8 casos de uso: youtube.com/watch?v=x8KP4Mdv… Cómo lo usa alguien que ya trabaja con esto: youtube.com/watch?v=ugILNhPo… Grok Bot versus Hermes: youtube.com/watch?v=t7aBvFxO… OpenClaw versus Hermes versus Grok Bot, la comparación más clara: teapawt.org/_avichawla/status/2090… dailydoseofds.com/p/grok-bot… Documentación oficial, esto es la base: docs.x.ai/grok-bot/get-start… docs.x.ai/grok-bot/bots docs.x.ai/grok-bot/skills-ro… docs.x.ai/grok-bot/computer-… x.ai/news/introducing-grok-b… Mi post del desk, 10 agentes, 2 días, paper de 50 mil dólares: teapawt.org/MatiasScalbi/status/20… Si este post les sirve, el lunes armo el tutorial del Desk de inversiones desde 0: qué bots, qué pedidos, Alpaca paper, Discord, y cómo probarlo con dinero demo Recuerden, es todo aprendizaje y no hay nada mejor que embarrarse HACIENDO con estas nuevas tecnologías Si yo pude, que no soy ningún profesional y hace 5 meses no sabía nada de IA, vos también podes Abrazo a todos y a meterle
6
4
75
7,381