More projects built with Dots
Continue exploring projects, experiments, and ideas from the Dots community.
ChatGPT Dots: How to Use OpenAI's New AI Agent
I spent my first hour with ChatGPT Dots, OpenAI’s new AI agent, and it got way more useful way faster than I expected. I gave it access to my calendar and a computer, then started building actual workflows around my day. In the first hour it helped organize 100+ calendar entries, fixed bugs in my iOS app, set up recurring finance/work tasks, and I could even call it like a normal phone conversation. In this video I walk through how to set up ChatGPT Dots, what it can actually do, cloud computer vs. local computer access, automations, calendar workflows, finances, coding, and what I’m planning to build next. I’ve been messing with AI agents like OpenClaw for almost a year, and Dots feels like one of the first versions that normal people could actually set up and use. Parked Disc Golf: https://parkeddiscgolf.net/ Download Parked Disc Golf on iPhone: https://apps.apple.com/us/app/parked-disc-golf/id6759809071 TIMESTAMPS 0:00 What ChatGPT Dots did in my first hour 0:53 What Dots actually are 1:28 How to set up ChatGPT Dots 2:36 Cloud computer vs. your own computer 3:45 The workflows I immediately automated 4:04 Using Dots to improve my iOS app 5:14 Connecting it to my finances 6:31 Calendar management + call prep 8:18 My Telegram / trading assistant idea 8:49 What I built in one hour 9:09 First impressions If you’re into ChatGPT, OpenAI, AI agents, automation, vibe coding, Codex, or personal AI assistants, subscribe. I’ll keep documenting what I build with this. #chatgpt #openaiDot "Dobby" booked a birthday lunch and compared flights
I've been trying out Dots from ChatGPT, and I'm seriously impressed. So far, I like it better than Muse. I named mine Dobby, and today he compared flights, booked a birthday lunch for my mom, and caught a work deadline I needed to keep an eye on. Having an AI assistant that can actually get things done is pretty exciting.
My Dot fixed a software bug and a data issue overnight while I slept
I went to bed with a software bug and a data issue. I woke up to both resolved. 🤖😴 While I slept, my Dot worked through the assignments, tested and verified the fixes, and left me a morning report. It also identified and tested a possible product enhancement—but stopped short of deploying it without my approval. That last part may be what impressed me most. Is this the new way we’re going to work—humans setting direction while AI keeps the project moving through the night? 🤔🤖
Dobby the Dot: flights compared, lunch booked, deadline caught
I’ve been trying out Dots from ChatGPT, and I’m seriously impressed. So far, I like it better than Muse. I named mine Dobby, and today he compared flights, booked a birthday lunch for my mom, and caught a work deadline I needed to keep an eye on. Having an AI assistant that can actually get things done is pretty exciting. 🤖
My Dot reconciled a budget discrepancy like a human coworker
Here's me giving my Dot the kind of direction a human coworker might get: "The answer should be somewhere in that spreadsheet." 🤖😅 I pointed it toward the budget file and its remittance section, but didn't tell it how to solve the problem. The AI found the relevant data, checked the fees against the Wise records, reconciled the discrepancy, and came back with the answer. Meanwhile, it was also finishing a separate calendar assignment I'd given it earlier. What I especially liked was that, after finding the missing context, it acknowledged that its earlier assumption had been wrong. Increasingly, this feels less like giving a computer instructions and more like giving a capable coworker a clue and letting them take it from there. 🤖🧐
Dot audits a codebase and plans a regression suite
我終於找到 dots 的正確打開姿勢了ww — the attached screenshot shows the creator's Dot delivering a 'Sent by Your dot' status report after 3m44s of autonomous work: the Dot verified the repo state of nurockplayer/tachiko-sheet (remote, main at f44ad233, uncommitted files preserved), summarized the live handoff (PR140/#138 done, production writer released, #142 DESIGN READY), read the project's delivery skill guidance, and drafted a minimal behavioral-regression plan — all without modifying files, creating worktrees, running tests, or pushing (waiting on the user's approval).
OpenAI Dots와 30분 통화하며 일 시켜본 결과 | 되는 것과 안 되는 것
✔️ ChatGPT에 새로 생긴 개인 에이전트 Dots, 전화를 걸어서 일을 맡기면 어디까지 해줄까요? 이번 영상에서는 제가 만든 봇 ‘싹쓰리봇’과 30분 동안 음성으로 통화하면서 직접 확인해봤습니다. Dots는 OpenAI가 DevDay 2026에서 발표한 기능으로, ChatGPT 안에서 계속 켜져 있는 개인 에이전트입니다. 이름과 캐릭터를 붙일 수 있고, 자기 전용 클라우드 컴퓨터와 브라우저로 일하기 때문에 대화가 끝나도 맡긴 일을 이어서 합니다. 통화하는 동안 맡겨본 일은 이렇습니다. 1. 20초짜리 Dots 소개 영상 제작 — 통화 도중 말을 끊고 Vox 스타일 종이 콜라주로 방향 바꾸기 2. 스페이스에 쌓이는 결과물 확인과 봇 꾸미기(움직이는 진돗개 아바타) 3. 매일 아침 7시, 최근 3일 메일 중 중요한 것만 요약해 채팅과 Slack으로 보내는 예약 작업 4. Canva·Bright Data 플러그인 설치 요청 — 로그인과 인증은 여전히 사람이 해야 하는 부분 5. 내 컴퓨터의 Claude Code와 봇의 원격 컴퓨터에 같은 쿠팡 상품 조사 시키기 잘 된 것만 보여드리지 않았습니다. 봇이 진행 상황을 잘못 보고했다가 스스로 정정한 장면, 쿠팡이 자동 접속을 막아 로컬과 원격 모두 조사하지 못한 장면도 자르지 않고 그대로 넣었습니다. 사용 한도 차감 기준과 쓸 수 있는 요금제는 본문 화면에 공식 원문으로 정리해뒀습니다. 여러 일을 한꺼번에 던져두고 전체 진행을 챙기게 하는 AI 비서가 궁금하셨다면, 실제로 써본 흐름을 따라가 보세요. 🔍 구독과 알림 설정을 하시면, 새로 나온 AI 에이전트를 직접 써본 후기를 가장 빠르게 받아보실 수 있습니다. --- 📍 타임라인 (Timeline) 00:00 | 인트로 01:57 | 1. Dots와 통화 시작하기 — 말을 끊고 끼어들기, 봇의 자기소개 04:18 | 2. 통화하면서 일 맡기기 — 20초 소개 영상 요청과 Vox 스타일 수정 07:48 | 3. 클라우드 컴퓨터와 프로필 중심 봇 — 진행 상황 묻기, 제어권, 선톡, Slack 11:55 | 4. 스페이스와 봇 꾸미기 — 결과물 저장소, 페이지·사이트·이미지, 아바타 16:10 | 5. 예약 작업 걸기 — 매일 아침 7시 메일 요약을 채팅과 Slack으로 18:32 | 6. 결과물 확인하기 — 20초 소개 영상 완성과 평가 22:04 | 7. 여러 일을 맡기는 팀장 봇 — 플러그인 설치 요청과 사람이 해야 하는 인증 25:54 | 8. 로컬 Claude Code와 원격 컴퓨터 — 쿠팡 조사 실험과 실패 원인 32:34 | 아웃트로 --- 🔗 관련 링크 및 참고 자료 1. OpenAI Dots 소개 (공식 발표, 한국어): https://openai.com/ko-KR/index/introducing-dots/ 2. Dots 사용 가이드 (시작하기·채널·작업과 기억·컴퓨터와 앱): https://learn.chatgpt.com/codex/dots 3. ChatGPT Pro 요금제 안내 (Dots 사용 대상 확인): https://help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers 4. DevDay 2026 발표 요약: https://openai.com/index/devday-2026-recap/ - 함께 보면 좋은 영상: (AI 팀메이트 플랫폼 · DeskRPG for Hermes): https://youtu.be/4NCXTWBxcN0 --- 💡 카카오 오픈 채팅방 [실전AI] 커뮤니티 https://open.kakao.com/o/gURfTmqh 🌐 단테랩스 홈페이지 https://dante-labs.com --- #OpenAI #Dots #OpenAIDots #DevDay2026 #ChatGPT #Codex #ClaudeCode #AI에이전트 #AI비서 #개인에이전트 #음성AI #AI자동화 #단테랩스
I Tried ChatGPT Dots: Setup, Voice Calls & Real Tasks
I set up my first ChatGPT dot, threw a bunch of tasks at it, called it twice, and my usage still showed 94% at the end of this test. Amazon blocked its browser. The DevDay infographic pulled in my branding without me asking. A pretty interesting first run. I named it Roger, found out it had a female voice, and renamed it Dolly. Then I gave it an infographic, an electricity research task, a Times Square restaurant search, and a cat toy search while I poked around its cloud computer. You get the setup, the calls, and the bits that worked or got stuck. I want to see how much work I can hand this thing over the next 30 days. The routine suggestions were useful: meeting prep, open-loop follow-ups, and filtering AI news. I'm going to try it on the work already taking up my day. ⚙️ TRAIN WITH ASPI https://goaspi.com 🧠 JOIN OPERATOR INTELLIGENCE https://www.skool.com/operator-intelligence ✍️ READ THE WRITTEN STUFF https://johnaspinallsr.substack.com 📬 EMAIL ME john@goaspi.com 0:00 OpenAI DevDay and ChatGPT dots 0:44 Where your dot lives: Codex vs ChatGPT 1:25 Creating a dot 2:14 Customizing the avatar 4:30 Roger's cloud computer 5:40 Calling my dot 7:12 Renaming it Dolly 8:29 The dot guide and limits 9:14 Firing off tasks in parallel 11:55 Amazon blocks the dot 13:09 Second call: what are you working on? 13:58 Results and the DevDay infographic 14:25 10 automations it suggested 15:15 Verdict
Dot reconciled her scattered event bookings and booked the rest
Kids aren't being fed breakfast this morning. Mum's busy trying out Dots!!! A friend texted me this morning saying, "You HAVE to try this, it's insane." Obviously, I did. I'm heading to Moree and Narrabri soon and a mate sent me a list of activities she'd booked while she's there. Problem was, I couldn't remember what I'd already booked, what I hadn't, or where any of the confirmations were. So I gave Dot her list and basically said: Go through my inbox. Work out what I've already booked. Tell me what I'm missing. Then help me book the rest. And it did. This is where agents start getting REALLY interesting. They're not just answering questions. They can go into the tools you already use, find information scattered everywhere, work out what needs doing and help you actually do it. If you're wondering what you'd use an AI agent for, think about all those jobs that start with: "I know that's in my inbox somewhere..." Thanks for ruining breakfast, @countrysmart.ai Right. Better go feed the kids.
My Dot 'Arnold' proactively flagged a business-process issue — with a fix ready
Помните мемы полуторагодовалой давности, когда GPT сам пишет пользователю и тот удивляется типа: это ты мне сейчас сам написал?! Так вот мем зашел слишком далеко. Арнольд в Дотс сам написал мне сегодня и подсветил нюанс в одном из бизнес-процессов, требующих внимание, с уже подготовленным им решением и вопросом применить ли это? А потом через пару часов маякнул мне о висяке в моем инбоксе, от которого косвенно и довольно неочевидно зависит другой важный и актуальный для меня вопрос.
Early-access verdict: Slack panes + phone calls with a Dot
I had early access to OpenAI's new "dots" and I have thoughts: 1. It's more work-focused than Grok Bot or muse... 2. It's great in Slack. Your team can message your bot directly and your bot can take actions, like spin up new panes. 3. You can "call" your dot... I loved being able to chat with it while I drive to work and my hands aren't available. 4. It should have been a separate app. 5. As a personal assistant, it's not as good as Grok Bot or Muse...yet. 6. Payments don't really work yet.
My Dot turned a call recording into a finished video — editing, subtitles, motion graphics, music, and YouTube upload
AI한테 이름 지어주고 전화해봤어 📞 OpenAI DevDay에서 발표한 dot. 나는 ‘닷율’이라고 이름 붙였어 ㅋㅋ “바이브메이킹 1기 학생들한테 인사해줘” “너 디스패치랑 뭐가 달라?” “통화 내용 자세히 정리해서 남겨줘” 직접 물어보고, 안 되는 것도 확인해봤어. 그리고 이 통화 녹화본을 닷율한테 주고 편집·자막·모션그래픽·음악까지 같이 만들었음. 유튜브 업로드도 맡겼고. 물론 한 번에 나온 건 아니야. 정렬 이상하다, 목소리 잘렸다, 모션이 부족하다… 내가 보고 피드백하면서 계속 고쳤어. 이 과정까지가 내가 AI를 쓰는 방식이거든. 닷율 첫 통화, 87초로 담아봤어 👇 https://www.youtube.com/shorts/4Bl9jJ1hYVU
ChatGPT Dots are SO Cool!
While I was flying, my dot spotted a sponsor workshop clashing with judging at my planned 1,000-person Delhi hackathon. Extending the day added 15Lacs and a later finish. A smaller workshop cost 5Lacs, but capped attendance at 250 and used participants' build time. My dot weighed those trade-offs against my budget and judging commitments, then prepared an updated sponsor offer, timetable and 10Lacs budget. I approved sending the proposal. Sponsor confirmation remained pending. That's the planning work dots by @OpenAI helped take off my plate.
OpenAI dots深度实测：Godot贪吃蛇与Blender三维模型并行开发
OpenAI dots deep test (Sept 30, 2026): I had my Dot develop a Godot snake game and a Blender Stonehenge 3D model in parallel on its cloud computer — got the source code back, played the game myself, and verified the model files and render images. Then I had it install and call Grok CLI to make a simple login-page demo. I also tested voice workflows: finding old notes in Notion, searching papers, then writing the organized content back to Notion. These demo cases were simple games, models and pages; the scheduled-task part showed only the setup flow, not a long unattended run.
I Tested OpenAI Dots, the AI That Never Sleeps
OpenAI's new Dots are always-on AI agents: each one runs on GPT-6 Astra, gets its own cloud computer and browser, and works across 4,000+ apps. I got one and named it Jarvis. It couldn't tell me what model it runs on, found today's top AI news in under a minute, and built a working habit-tracker app in about 10 minutes. The catch: 20 MB uploads in chat, no PC access on my Pro plan yet, and a lot of stalling. Tests run on my own Dot on Sept 30, 2026.
OpenAI Dots Review: It Found $450 I Forgot
I gave OpenAI's new Dots AI agent one real job: find the money hiding in my inbox (a test inbox I set up). It found an unpaid invoice and drafted the follow-up while I was away. Next I tested it on my content and on fake emails built to trick it. Full review: https://youtu.be/exLsCMTm-Vw
Madiyar's Dot audits his calendar, inbox, Slack, and Codex agents on its own
Dot от OpenAI супер классная штука. Сам анализирует мой календарь, почту, слак, проверяет моих агентов в codex, говорит что там не так, создаёт тикеты, спрашивает что делегировать другим агентам. Это уже другой уровень, что приближает нас точно к AGI. https://openai.com/index/introducing-dots/
