← Back to leaderboard
arceyul

arc.

Verifiedactive
@arceyul

ai marketer info@aiscwork.com

RosterAI Agents & Productivity ToolsStartup & SaaS Builders
111,460 followers · 1,051 following · Last polled 3h ago
Perf. Score
48.2
Rank #248
Value Score
24.8
#213 of 273 priced
Median views
2.4K
p25 2.2K · trailing 7d
Followers
111K
-0.1% 7d
ER (impr.)
2.11%
eng ÷ impressions
ER (foll.)
0.12%
eng ÷ followers
Est. CPM
$209.12
QT-rate basis
WoW views
-71.8%
▼ falling

Rates & value · est. from 2.4K median organic views

overpriced
Quote tweetbasis
$500
$209.12 / 1K views
Post
—
no CPM
Thread
—
no CPM
Retweet
—
no CPM
Value Score 24.8 · rank #213 of 273 priced$3194.89 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (2.4K), not the mean (6.3K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (2.2K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
https://t.co/DSYvgrk1Isfrozen19d ago382K6275427121.2K0.50%
SIGAN A ESTA PERSONA NO DIGO NADA MÁSfrozen14d ago232K1.2K1051401.3K1.11%
SIGAN A ESTA CUENTA NO DIRÉ NADA MÁS…frozen
11d ago
159K
1.8K
123
6
0
1.9K
2.35%
🚨 ¿QUÉ ESTÁ PASANDO EN TIKTOK? un personaje UGC hecho con IA, con solo 7 vídeos, tiene 219K seguidores y 49,6M de visualizaciones los números: - 7 publicaciones, todas por encima de 2M de views - ~7M de views de media por vídeo - 3,8M de likes, un 7,7% de like rate - 0 seguidos, sin caras, sin hablar cómo montar tu propia versión: - paso 1: diseña un personaje en GPT Image 2.5 con un look inconfundible (pelo, bigote, traje) - paso 2: actúa tú mismo el movimiento o descríbelo beat a beat - paso 3: Seedance 2.5 mete a tu personaje en tu escena usando la imagen del personaje como referencia dentro de Higgsfield AI Influencer - paso 4: mismo personaje, misma cara de póker, localización nueva en cada post cómo llegar a 10K$ con esto: - las views por sí solas pagan poco, el dinero está en lo que pones detrás - 2 o 3 marcas pagando por integrar su producto en la escena del personaje - un link en bio a tu propio producto o a un afiliado - o vender el personaje como servicio: 4 o 5 marcas que quieran el suyo ya suman 10K$ la parte de IA es la fácil. el diseño del personaje es lo que consiguió 49,6M de views GUÁRDATE ESTOfrozen18d ago80K638972476971.83%
1B ARR in less than 2 years, 30M users and that’s only less than 0,5% of world population… imagine when AI goes mass adopted, Higgsfield is competing with industry giants and with way LESS time here CRAZYfrozen28d ago57K222156170250.73%
🚨 WHY IS NOBODY TALKING ABOUT THE REAL AI INFLUENCER BUSINESS? everyone's trying to get followers. the money is somewhere else. the real business: → you don't sell the influencer → you sell the ads it makes → to brands that already pay for creatives every month how to set it up: - step 1: build your character in Higgsfield AI Influencer - step 2: pick a niche: skincare, beauty, perfume, fitness - step 3: recreate the UGC formats that convert there with Genjutsu - step 4: make 3 samples and DM 10 DTC brands why brands say yes: → their ads burn out fast → UGC creators are slow and limited → your character works 24 hours/day one character. any niche. a new batch of ads every month. 🔖 SAVE THISfrozen17d ago46K175950180.45%
Before: you ask Claude Code something, copy the answer, paste it into Cursor, then explain it all again to Codex. After: they're all in one group chat with you, talking to each other. That's what Bolter does. It's a messenger for people and AI agents, and your agents keep running wherever they already run. Want to bring in a friend's agent? Their owner approves it first, so nothing connects without a yes. It even works from your phone, and connecting your own agents is free. https://t.co/FfqHC50ND1frozen9d ago46K126213020.31%
SI PROGRAMAS CON CLAUDE CODE MIRA ESTE ANUNCIO… NO DIRÉ NADA MÁSfrozen23d ago45K25911102351.12%
SIGAN A ESTA PERSONA NO DIRÉ NADA MÁS…frozen9d ago41K1614070280.58%
this bubble has been popping for at least 3 years nowfrozen15d ago38K11021020.30%
any video is a template now 👀frozen14d ago34K14046070.47%
🚨AHORA PUEDES DEJARLE TU CORREO COMPLETAMENTE A LA IA Inbox Zero es open source y escribe los emails con tu propio estilo, como si los hubieras redactado tú. Junta todas tus cuentas de Gmail y Outlook en una sola bandeja y la divide para que lo importante quede siempre arriba. Además te da de baja de todo el spam de una vez. Funciona en web, Mac, Windows e iPhone. Y como el código es abierto, puedes comprobar qué hace exactamente con tu correo. GUÁRDATE ESTOfrozen5d ago33K1421790360.62%
“yo ChatGPT run all my ads for my marketing agency. make no mistakes.” https://t.co/m9seVLZbkvfrozen17d ago31K2739040.14%
THIS IS CRAZY Someone built an entire marketing team with Grok 4.7. 5 AI bots. 1 human. 20M+ views. The Researcher digs through competitors while he sleeps. The Creative Director turns the best idea into a concept. The UGC Creator makes it with an AI avatar that looks completely real. The Publisher drops everything into Postiz. One of those posts hit 7.5M views and brought in 3.3K followers. No actors. No agency. No hours lost scheduling. He just broke down the entire setup step by step:frozen28d ago27K14935080.61%
be Yash Daftary: > sells event merch with two friends as a teen > starts a protein gum company that never makes a dollar > builds a rental marketplace, pitches it at 80+ competitions > VC analyst at 20, running Facebook ads from his Emory dorm > starts FanBasis during COVID > Forbes 30 Under 30 > rebrands to Commas now he's on the other side of the table: teaming up with @simonsquibb on the world's biggest business competition, to help the next founders build theirs.2d ago27K16535050.66%
we thought we would not see any resets BUT GUESS WHAT THEY JUST DID https://t.co/xL15iAPfl7frozen24d ago26K13512020.53%
omg SI looks terrible. AI is just objectively better and 99% of people agree lolfrozen22d ago26K5513010.23%
Tavus gave me early access to Griffin, and it reframed how I think about AI. We've been treating it like it only lives in text. But here it watches your hands, follows a Rubik's cube move by move, and solves it with you. That's participation, not answering.frozen21d ago24K11221010.49%
I used to fix AI voiceovers by generating the same line in five tabs until one of them said the name right. @OnepinAI listens for clarity and pronunciation on every take and fixes the wrong words itself. No more tab roulette.frozen21d ago24K7610020.33%
Agents don't struggle with CRMs because the CRM is bad. 
They struggle because they can't query it directly. Attio fixed that today with a SQL engine inside the CRM. 
→ data connectors pull in the rest of your stack
→ up to 100M records per object
→ activities and interactions now in beta Nothing in the middle. No warehouse, no BI tool, no analyst. Your agent gets answers straight from the pipeline.frozen8d ago23K11736010.54%
the best thing an AI did for me this week was ask a question.Gamma 5 is out today, and the agent doesn't build until it understands the job: -who is this for? - what should it lead with? - what's your setup? - how long? answer those and you get the full deck, built around your answers.skip one, like i did, and it leaves it alone instead of making stuff up. that's the difference between a tool that guesses and one that listens. your old decks stay saved, and Gamma Classic is one click away.frozen16d ago23K16821010.73%
Most AI assistants today: "Here are 5 steps to export your reports." Pine Computer: *logs in, exports them, zips them, done* That's the difference between an AI that answers and an AI that actually works. #PineComputer #ComputerUse #DeveloperTools https://t.co/m0PRXOEtR1frozen13d ago23K9632040.46%
every AI chat ends the same way: you get an answer, you close the tab, and whatever you made disappears into the scroll. Eazo does it differently. you talk to the agent only to create. you describe the app you want, it builds it, and from then on you just open it and use it like any other app. no prompting it again every time you need it. describe → build → open → use. the conversation is how you make the tool. it's not where the tool has to live. #eazo #personalagent #AIAgentsfrozen14d ago22K11622020.55%
I thought AI can’t replicate this level of motion design. Then I tried higgsfield’s new Katana💀 https://t.co/t0JOMsCtKtfrozen8d ago20K6437070.40%
These are publicly accessible website, humans access them all the time. What’s the difference here?frozen26d ago20K1611000.09%
OpenAI Dots × Higgsfield looks CRAZY trying this asapfrozen23d ago19K583120210.49%
>shot everything in log because a >youtuber said it looks “cinematic” >footage comes out grey and dead >open davinci >close davinci >8 months of grey footage sitting on my drive now experimenting with tesseract by @trymirage to give my agent the tools to actually color-grade footage based on what i’m going for might finally get around to fixing those shotsfrozen8d ago18K12821010.74%
🚨 Este tipo acaba de FILTRAR el metodo MAS FACIL para hacer dinero en 2026 Claude Code + YouTube = $62,000/mes Nada complicado (Solo 13 minutos) Guarda esto en favoritos para que no lo pierdas https://t.co/GWmIiCboFhfrozen24d ago15K23535203063.90%
EL GRUPO DEL VIAJE TIENE 500 MENSAJES. 40 SON MEMES Y 200 SON "¿Y AL FINAL QUÉ?". RENE LO CIERRA EN UNA TARDE. fechas, presupuesto, quién se apunta, quién se raja. el grupo se llena y nadie decide nada. añades a Rene como un contacto más, sin que nadie instale nada. le pides que resuma lo que habéis hablado y en dos líneas tenéis claro dónde estáis. luego os propone tres opciones según el presupuesto, reserva la más votada, divide los gastos y os deja el calendario del viaje listo. todo pasa en el grupo, a la vista de todos. lo privado se queda en privado. de "¿y al final qué?" a billetes comprados. sin apps, solo escríbele. @Rene_Labsfrozen10d ago13K14445151.23%
so they really just fired the internal watchdogs huhfrozen21d ago13K8422010.71%
Voice typing is useful for getting words onto a screen. A personal agent has a different challenge: figuring out what those words are asking it to do. That's why I'm interested in how SpeakON approaches everyday requests like reminders, email, and scheduling especially when a request needs more context.2d ago11K15733031.47%
GitHub acaba de solucionar el mayor problema del vibe coding. Acaban de lanzar Spec Kit y en días ya tiene +95K estrellas. ¿La idea? En vez de tirar prompts vagos y rezar para que el agente no rompa tu proyecto… Spec Kit obliga a la IA a crear una especificación estructurada ANTES de tocar código.frozen18d ago6.4K10410140993.54%
Be Sujal Dhakal > grow up in a small village in Nepal
> move to America without knowing what startups are
> dad becomes a disabled veteran, mom’s autoimmune disease gets worse
> take on a second job coordinating healthcare for your parents
> start building the help they needed at Georgia Tech
> drop out to work on Vene full-time
> get backing from tier 1 VCs in SF
> already deploying Vene to 50k patients across the US Vene helps schedule follow-up appointments, get records to the right doctor, and handle the paperwork going back and forth with insurers:8h ago6.3K9013091.63%
SIGAN A ESTA CUENTA NO DIRÉ NADA MÁS…2d ago5.8K1931950334.30%
extremely important read for engineering leadership:frozen28d ago5.8K144106032.83%
los que no quisieron estudiar informática, viendo que ahora el software con IA se cura y mejora solo: https://t.co/7TJvwAAd1c1d ago4.8K185460114.25%
TU IA NO ESCRIBE COMO TÚ PORQUE NUNCA HA LEÍDO NADA TUYO. Le pides un post y te devuelve algo que podría haber escrito cualquiera. Normal, empieza de cero cada vez. Lo que hice: → Conecté a Brain mis notas, guiones y posts antiguos → Abrí Claude → Le pedí ideas nuevas con mi estilo → Me las dio citando los posts míos en los que se basaba Ya no le explico cómo escribo. Lo ha leído todo. 🔖 Guárdalo y dedícale unas horas.frozen29d ago4.5K374140151.54%
🚨Claude hizo TODO lo de este clip. no abrí ningún dashboard. no toqué ningún botón. le dije lo que quería y lo hizo. lo que usó por debajo es Zernio vía MCP: > 16 redes desde una sola interfaz
> WhatsApp, iMessage, SMS y llamadas en el mismo sitio
> anuncios en 7 redes y aquí viene lo que te afecta aunque no sepas programar. piensa en cuántas horas de tu semana se van en abrir una app, buscar el menú, rellenar campos y darle a guardar. eso NO es tu trabajo. es el peaje que pagas para que la herramienta te entienda. ese peaje ya no hace falta. ahora le dices a la IA lo que quieres con tus palabras y ella se encarga de hablar con la herramienta. en tu día a día se ve así: → "programa este post en todas las redes para mañana a las 9"
→ "contesta a los clientes que escribieron por WhatsApp sobre el pedido"
→ "pausa los anuncios que no estén funcionando" tú decides qué hay que hacer. la IA hace los clics. para empezar, apunta la tarea que más repites cada semana y pídesela a la IA en una frase. con una sola que te quites de encima ya lo notas. después de una tarde así, volver a hacerlo a mano se siente como rellenar un formulario en papel. GUÁRDATE ESTOfrozen17d ago4.4K67940432.78%
sending this to my unemployed friends ASAP!frozen20d ago4.0K2702040.83%
everyone's obsessed with bigger models. almost nobody is talking about memory. Volantis uses optics to move data farther across the package than electrical interconnects can, scaling memory bandwidth and capacity by orders of magnitude. that could open a different hardware path for AI.frozen21d ago3.8K16960004.65%
SIGAN A ESTA CUENTA NO DIGO NADA MÁS…frozen22d ago3.6K2210070.83%
Knowing which pages bring my competitors customers always meant paying for Similarweb on top of everything else, so I never did it. going to try the competitor breakdown on SpringBrand this week. curious how close it getsfrozen24d ago3.6K14922014.27%
consumer brands NEED thisfrozen17d ago3.5K600010.20%
Hold on, you can rewrite an actor's expression after the shoot is over? DRAMA is auto-tune, but for performances. Choose the emotion. Dial in how strong. Hit generate. The face, the lighting, the camera move all stay exactly the same. It looks like cinema, not AI slop. https://t.co/VOKYBUTqK8frozen14d ago3.5K1372110114.67%
this account shares the most useful enterprise AI content on X right now.frozen21d ago3.4K610000.20%
lmaooo someone do a series of this ASAPfrozen15d ago3.4K2312040.88%
RIP to every startup whose product was "a nicer dashboard" https://t.co/f9AERk6o2zfrozen7d ago3.2K10622003.47%
this is terrifying… https://t.co/pfa66japLk2d ago3.2K17026035.72%
back in Spain after 3 weeks in Bali. lived it and loved it. now back to what matters, work work work.frozen21d ago3.2K3541071.49%
FINALLY SOMEONE BUILT THIS Alook puts your coding agents in the same chat as YOUR TEAM. Open source, running on your own machine. save this https://t.co/nB9UJfWPgyfrozen13d ago3.1K5835052.26%
YOU CAN NOW EDIT PREMIERE AND AFTER EFFECTS FILES WITHOUT OPENING ADOBE Until now, if you wanted your agent to edit a .prproj, it had to drive Adobe like a human. Before: → Open Premiere → Click through the UI with computer use → Save it back as a .prproj Now: → Open Claude or ChatGPT → Say what you want changed → Get the edited project, rendered Tesseract works directly on supported .prproj and .aep files. No Adobe at all. Bye, Adobe.frozen15d ago3.1K9721023.28%
$50M to read your mind. Literally Sabi is building a cap that picks up brain signals through neuroimaging sensors and converts them into prompts WHAT IS THISfrozen13d ago3.1K5210001.73%
I made $20,000 last month. Never showed my face. Never hired a creator. Never touched a camera. Just 40 videos. A $20 Claude subscription. Postiz to schedule and post everything. And a workflow that most people still don't know exists. 3 of those videos pulled 2 million views combined. Back in the day, the biggest barrier to UGC was finding the right creator, negotiating rates, waiting on revisions, hoping they delivered. Not anymore. 40 videos. A few hours of work. $20,000. Here's the exact five-step build I used, from the first script to Postiz publishing it all for me.15d ago3.1K188550136.90%
honestly the "share it like a google doc" part is what got mefrozen15d ago2.7K9611033.79%
IF YOU WORK AT A STARTUP, THIS SAVES YOU HOURS EVERY WEEK this tool kills the three tasks that eat the most time: → answering the same question twice → chasing follow-ups → updating the docs you give it your company's context (customers, conversations, docs, projects and CRM) and an AI agent handles it. it's open source: self-host it free or let them run it for you.frozen14d ago2.6K130760115.90%
THE TRICK TO PAYING LESS FOR CLAUDE AND GPT THAT ALMOST NOBODY USES: AntSeed Gateway connects your agents to an open market of 700+ models. Providers compete, and every request goes to the cheapest verified one. Each agent gets its own API key and spend limit. Set up yoursfrozen9d ago2.6K136121035.88%
anyone knows why is that?2d ago2.6K1201010.55%
Every AI video tool makes the same trade: more creative freedom, less control over who's on screen. P-Video-2-Pro Reference from @PrunaAI goes the other way. → up to 9 reference images, 3 videos and 3 audio clips
→ same character, product or voice in every clip
→ 5 to 15 seconds at 480p or 768p, 24fps, sound included
→ built on MiniMax H3 Ref2V Cost mode starts at $0.02/s. Speed mode generates in as little as 1.9s. Live on the API today. https://t.co/TtSCg8jlpofrozen7d ago2.6K9322013.83%
an agent just signed up to your product. can you answer these? → who is it? → who owns it? → is it the same one that logged in yesterday? → who do you call when it breaks something? most apps can't answer any of them. AgentID answers all four. it works like "Sign in with Google" for AI agents: each one gets its own identity and email, linked to a real person. it's standard OIDC, so it plugs into Clerk, Auth0 and Supabase. free for apps.frozen16d ago2.5K902010.49%
Bolt is free this weekend. Type an idea, get a full-stack web or mobile app in your browser, and own the code. Best project by Oct 20 takes home $10k.frozen5d ago2.4K13426005.86%
CHESS ENGINES SOLVED THIS DECADES AGO. LLMs ARE ONLY GETTING IT NOW. A chess engine never says "sorry, out of time". You stop it, and it plays the best move it's found so far. That's called an anytime algorithm. Old idea. Almost nobody applies it to LLM inference. Today, if a reasoning model hits your timeout, you get nothing. Hit the token limit and you get half a thought. Span-02 works like the chess engine: Set a deadline → it stops reasoning → you get its best answer so far + its confidence. Interruptible reasoning, not a timeout that fails. Feels obvious once you see it. That's usually a good sign.frozen7d ago2.4K9424004.24%
they cooking over there oh2d ago2.3K15521026.95%
I used to regenerate entire AI voiceovers over a single mispronounced word. Then I tried Drama 3 from @FishAudio. → select the word
→ change the text, tone or emphasis
→ everything else stays put What used to take a dozen retries now takes one click. AI audio is finally editable like text.frozen9d ago2.3K12030005.44%
shipping solo just got way easier. synteq launched apps and functions. you deploy with one command, a github push, or by dragging a folder into https://t.co/OPm1GuCOUs. no servers, no nginx, no ssl pain. you get a live url, health checks with zero downtime and one-step rollbacks. comment "LAUNCH" under their post for $20 free credit ↓2d ago2.3K15432037.19%
LAUNCH VIDEO OF THE YEAR, LOVE THE VIBE these are real actors right?frozen7d ago2.2K8922004.17%
Most coding benchmarks test the same handful of languages. With a compute grant, that's the first gap I'd close: an open eval set for coding agents in the languages everyone else skips. Nebius is making that kind of project possible. Its AI Builder Program is coming to Europe with $10M in compute grants, capital and community support. Grants run on Nebius Token Factory, for European teams working on: 
→ open models
→ coding agents
→ inference tooling3d ago2.2K14923006.99%
i like when things like this pop up in my timeline definitely trying it 👀2d ago2.2K15431017.28%
No guidebook explains Rome in paragraphs. It shows you photos, maps and a plan for each day. Ask an AI for 3 days in Rome, though, and you get a 1,500-word essay. OpenUI 1.0 fixes that. The AI still does the research, but you get a real interface to decide with: cards, maps and buttons instead of paragraphs.frozen7d ago2.0K8830004.63%
I’ve been saying it for a while: world models are the next frontier after LLMs. Seeing one built specifically for engineering this early is a big deal. SKY-001 from Skyfall AI handles CAD and BIM, matches GPT-6 Astra on 3D construction accuracy at roughly 150x lower cost, and runs over 100x faster on BIM.2d ago1.7K6710014.01%
ACABO DE ENCONTRAR ESTA CUENTA Y ME SIENTO ASÍ https://t.co/3Wr7HyAP1r4h ago1.4K2398063.26%
LA IA YA ESCRIBE TU CÓDIGO En vez de lanzar a producción y rezar, Breken monta una copia realista de producción de tu app y la rompe a propósito antes de que llegue a tus usuarios. Crea usuarios y datos con historial, tumba servicios a mitad de tarea, prueba los órdenes raros, arregla lo que falla y lo vuelve a probar en cada release. Menos cobros duplicados y menos noches apagando fuegos. Funciona con APIs, apps y agentes de IA. 7 días gratis con GitHub, sin tarjeta6h ago1.1K801010.89%