Sponsored
$5/week ↗

More projects built with Dots

Continue exploring projects, experiments, and ideas from the Dots community.

Puse a mi agente de IA a vender webs de 197 $… y esto pasó

Puse a mi agente de IA a vender webs de 197 $… y esto pasó

Aprende a usar ChatGPT como un pro conmigo 1:1 → https://carlosrodera.com/chatgpt-session ¿Puede un agente de IA hacer trabajo comercial real? En este vídeo pongo a Alfred, mi Dot de ChatGPT, a trabajar en una misión concreta: buscar negocios sin página web, coordinar otros modelos, crear cinco demos y preparar una oferta de 197 dólares para contactar con ellos. Las primeras webs son demasiado básicas. Literalmente: «esto no lo compra nadie». Así que detengo el proceso, cambio el modelo, aumento el criterio de calidad y le pido rehacerlas con un enfoque más útil para el negocio. Después llamo a Alfred por teléfono para que me resuma nuestras primeras 36 horas trabajando juntos. Alfred reporta 158 emails enviados y 11 conversaciones comerciales con respuesta. Estas cifras son su propio balance verbal, no una auditoría independiente. En la prueba concreta de las webs se envía el primer email, pero no hay ninguna venta confirmada durante el directo. Al finalizar, Alfred informa de cinco rediseños preparados: una nueva versión publicada y cuatro enlaces todavía pendientes de actualizar. El carrusel para Instagram sí se crea y publica en directo. Lo interesante no es vender una autonomía ficticia. Es ver qué puede ejecutar un agente cuando tiene una misión, herramientas y permisos, y dónde todavía necesita criterio humano: - elegir el modelo adecuado; - corregir resultados mediocres; - proteger datos sensibles; - revisar permisos; - comprobar lo que realmente se ha completado; - cerrar las conversaciones comerciales. CAPÍTULOS 00:00 Esto no lo compra nadie 00:23 El plan 01:38 Alfred empieza a orquestar 05:46 Las primeras webs no sirven 07:16 Alfred me llama por teléfono 09:58 Esto ya es otra cosa 12:15 Crea y publica un carrusel en directo 15:05 Le delego nuevas tareas 17:11 Lo que he aprendido usando Alfred 20:01 Balance final RECURSOS MENCIONADOS Dots: https://openai.com/index/introducing-dots/ OpenClaw: https://openclaw.ai/ Jev / Typesafe: https://typesafe.ai/ La voz de Alfred que aparece en el vídeo pertenece al propio agente. Carlos Rodera AI Builder · Founder of Vocoding #InteligenciaArtificial #AgentesIA #ChatGPT #IAParaNegocios #Dots
@CarlosRodera
Reddit · Projects

My Dot finally automated YNAB transaction categorisation for me

I waited years for YNAB to automate categorisation. Today I stopped waiting. I’ve spent years wanting YNAB to take more of the manual categorisation off my hands. Every month or so, I download bank transactions, import the OFX files and work through the backlog. Today I had ChatGPT Dots work through it with me, using my existing YNAB budget and browser. It’s made me realise we don’t need to wait for YNAB to build every improvement we want. Here’s what we’ve done so far: - Checked my account histories, worked out the missing date ranges, downloaded the bank exports and imported 70 new transactions, matching another nine. - Categorised 297 transactions using my existing categories and past choices, asking me about purchases whose purpose wasn’t clear. - Opened my Amazon order history and matched 30 charges to the actual purchases. A mixed order was split between categories using the itemised invoice. - Matched 69 internal transfer pairs across my accounts, without changing their balances. - Traced a stubborn $6.50 discrepancy back to a missing coffee purchase from December last year. Once I approved adding it, the account balanced. The Amazon example really stood out. “Amazon” tells you almost nothing about what you bought. Looking at the order, identifying the items and applying my categories is the kind of automatic categorisation I’ve wanted for years. It still needs supervision. I’ve answered questions about work versus personal spending and gifts, and there are unresolved transactions left. It also takes time and needs the browser left free while it works. This wasn’t a one-click cleanup. But this level of analysis was not possible previously. But the shift feels significant: I can ask for the workflow I want in plain language, rather than wait for it to arrive on YNAB’s roadmap; which I’ve waited years for! YNAB remains my budgeting system. I no longer need every improvement to come from YNAB.
ChatGPT Dots Review: The AI Personal Assistant Replacing Notion?

ChatGPT Dots Review: The AI Personal Assistant Replacing Notion?

ChatGPT Dots Review: The AI Personal Assistant Replacing Notion? Can ChatGPT dots replace Notion as my daily planning system? Here’s my honest review of OpenAI’s AI personal assistant: the real workflows I’m using, why context matters, and the bugs I’m willing to wait out. I’ve been using my dot to turn brain dumps into priorities, stay focused on three daily goals, and follow through on what I say matters. In this video, I share how it’s helping with calendar planning, Gmail and delivery tracking, Uber Eats, Airbnb logistics, and monitoring replies on X. What I cover: • Why dots feels like a personal assistant that actually “up-manages” me • Moving from Notion planning to an agent that prompts me to act • Brain dumps, daily goals, and reducing context switching • Real examples across Gmail, my calendar, Uber Eats, Airbnb, and X • The role of memory, connected apps, and personal context • My experience compared with OpenClaw and Hermes • Phone-call bugs, voice limitations, and what I want improved My biggest hope for AI personal assistants: reducing the gap between our intentions and our behavior. This is my experience with dots at the time of recording, including what’s working and what still feels buggy. This video is not sponsored. Are you using ChatGPT dots, OpenClaw, or Hermes? Tell me which workflow has actually made your day easier. Subscribe for more practical AI workflows and conversations about frontier technology. #ChatGPT #AIAssistant #AIProductivity
@AsheMagalhaes
I Phoned My AI from the Grocery Store | OpenAI Dots & Ava

I Phoned My AI from the Grocery Store | OpenAI Dots & Ava

I Phoned My AI from the Grocery Store | OpenAI Dots & Ava I called my AI from the grocery store to steer an animation project. Is that the moment an assistant becomes a project manager? Brian and Veya discuss his experience using Dot Ava to keep local animation work moving for Ash’s Diary, a story drawn from his tabletop role-playing campaign. Brian explains the GPU-job babysitting he wants to hand off, why constant monitoring concerns him, and why he still reviews the creative result. The conversation also covers his reported fantasy football help, developing ideas and names for a space game, and a local scene gallery where he can approve a version or return specific revision notes. Veya asks the useful question: does adding projects free more of Brian’s day, or fill it with Ava’s questions? This is Brian’s experience and a discussion with an AI cohost, recorded October 4, 2026. Animated concept scenes are illustrations. Ash’s Diary excerpts show existing project footage; this recording does not establish Ava’s authorship of those excerpts. No measured token savings, personal cost comparison, speed benchmark, or guaranteed outcome is claimed. The ComfyUI context comes from Brian’s earlier project account; this episode is not a live ComfyUI demonstration. Like this video and subscribe to Human Behind the Machine for the follow-up on whether Ava makes these projects more efficient. Tell us: what ongoing job would you hand your dot, and which decisions would you keep? CHAPTERS 0:00 A production meeting at the grocery store 0:12 Human Behind the Machine — Dot Ava 0:19 Assistant or project manager? 1:14 Keeping local animation moving 2:10 Monitoring costs, boundaries and final approval 4:29 Ash’s Diary and the GPU queue 5:41 Calling Ava from the grocery store 6:36 Who is managing whom? 6:55 Fantasy football and breaking news 7:46 Creative help for a space game 8:29 An alternative to the same bad idea 9:11 The scene review and revision loop 9:59 What would prove greater efficiency? Official background: https://learn.chatgpt.com/docs/dots https://learn.chatgpt.com/docs/dots/tasks-and-memory #HumanBehindTheMachine #OpenAIDots #AIAgents #AIAnimation #HumanInTheLoop
@HumanBehindtheMachine
Reddit · Projects

Dot finished a rental report for mum, then emailed wife and co-ordered replacement printer filament with zero input

This morning I was just super disappointed with OpenAI. I haven't moved a long way from that, but my experience with my dot has been something that's worth talking about. I think I understand a bit more about what dots are for. When I first started using my dot, I used it as an orchestrator, like I would have previously on coding tasks. It struggled with anything complex. However, as the day went on, I started to change my use case of it, and I stopped using it as an orchestrator and more like a personal assistant. That was when I had a few aha moments. The first moment was when it emailed my mum and introduced itself... that gave me a bit of a shock. I told her I'd work on a report for rental housing she was looking for it. I wanted to give her daily report and she was struggling to find places. I'd done about half the job but not completed it, so I asked my dot to finish it off and send her the report. While doing that, it introduced itself to her and asked what else she needed. She sent me a really nice text message thanking me... Next, it identified a few things in my emails. Bambu Labs offered it to us free filament due to an issue we had with our A1 printer (power board exploded). The dot found this email which I'd forgotten about and I told it to ask my wife about it because I wasn't sure what the best choice was, and it emailed her. She had a few back-and-forth emails with it (while I was next to her on the couch laughing), and now together they've ordered the filament together, all with no input from me.