https://www.youtube.com/watch?v=Q9tF0R8d_Co John Aspinall builds a ChatGPT Dot from scratch on Pro: character + name, cloud computer, voice calls, parallel research jobs, an Amazon browser attempt, and routine ideas — with usage still at 94% at the end. Access And Create 1. Pro plan ($100) in this demo — Dot appeared under Your Dot after DevDay rollout lag 2. Prefer ChatGPT side over Codex tasks so Dot work does not burn Codex usage (he frames Dot usage as not hitting normal usage the same way) 3. Create Dot — always-on agent powered by Astra; setup spins up the cloud computer 4. Choose where it can work — Dot computer always on; optional toggle for your own computer 5. Customize character (presets or builder), colors, accessories; name it; save 6. UI color-matches the Dot once saved Cloud Computer 1. Open the Dot computer — Linux desktop with browser and terminal 2. Speed test in demo was modest vs what he had seen on Grok Bot cloud PCs 3. You can take control then return control to the Dot 4. Branded animations while it works (extra UI polish vs Grok Bot in his view) Voice Calls And Naming 1. Call the Dot from the phone/call UI — brief connection delay is normal 2. Voice matches your ChatGPT voice setting — changing Dot voice may change regular voice chats too 3. He named it Roger, heard a female voice, then renamed to Dolly rather than change voice globally 4. Dot confirmed GPT-6 Astra when asked; model picker is not exposed in the Dot chat the way normal chats are 5. Slack connector is available during setup (he selects Slack as an integration option) Parallel Tasks He Fired 1. One continuous Dot thread — no sub-chats / only one Dot for now (vs multiple Grok Bots) 2. DevDay infographic in chat 3. Research modern electricity / power over ~100 years 4. Top 10 restaurants for someone staying in Times Square 5. Safe self-summary of who he is / what he does (no client secrets) 6. List current connectors / what it can touch (email, computer, docs, Figma, Fathom, research)

https://www.youtube.com/watch?v=MhrH8oGVwiI 슈퍼휴먼X walks through a fresh ChatGPT Dots setup: create a Dot on Mac (Pro), call it on the phone, park results in Spaces, schedule a daily job, and pull the same agent into Slack. Access And Setup 1. Available on ChatGPT Pro in this demo 2. Windows not working yet for the creator — demo runs on Mac 3. Open Your Dots, create one, connect the computer 4. Set a name and character (pet-like shapes available), then save 5. Same features also showed up on smartphone after an update during filming Phone Call Demo 1. Tap the phone icon for a real voice call (text chat still works) 2. Ask the Dot to check calendar and email and list the top three things to do now 3. Dot pulled prior schedule priorities (multi-campus meeting, lecture prep) and offered to re-check email/calendar 4. Asked for a capability summary — Dot said it could handle five types of delegated tasks based on the work 5. Point of the feature: not just answers — checks email, calendar, Google Drive and acts like a sidekick assistant Spaces (Notion-Like Results Home) 1. Spaces store materials — framed as similar to Notion for projects, text, pages, images 2. Assigned: summarize recent OpenAI news and put it in the Space — later a page appeared with the result 3. Click into the page to organize like files/docs; paste images; edit like a document 4. Best mental model from the video: Spaces record conversation results, not the full chat transcript 5. Can collect Google Drive files plus your own images and data Scheduling Repeat Work 1. One-time tasks are fine; for every day / week / 24 hours use scheduling 2. Demo ask: summarize OpenAI news every day and save it to that Space — schedule applied 3. ChatGPT scheduling power applied to the Dot so the Space keeps filling without re-prompting Slack Integration 1. Connect Dot to Slack and add the agent as a team member 2. @mention the Dot and ask it to summarize — it organized recently uploaded materials in the channel

https://www.youtube.com/watch?v=nhb7mA4sZPw Javid Shaikh explains ChatGPT Dots simply: what a Dot is, the six parts that make it work, how you stay in control, and the risks before you hand it your inbox. What A Dot Is 1. Not a toy character — an always-on AI agent with its own cloud computer 2. You give a goal; it works in the background and brings results to review 3. Can juggle multiple tasks; you can name it; runs GPT-6 Astra under the hood 4. Unlike a normal ChatGPT tab that stops when you leave — a Dot keeps going Six Parts He Walks Through 1. Own cloud computer + browser (separate from yours; inspect / look over its shoulder) 2. App connections via OpenAI plugins (4,000+ apps) — connect fewer to reduce blast radius 3. Proactive research while you are away — framed as read-only background work 4. Memory / learning from feedback and notes over time 5. Reach you in ChatGPT (desktop/web/phone), with Slack / Teams testing and calls mentioned 6. Custom rules — what it can do alone, what needs approval, what it must never do Control And Safety 1. Sensitive actions (password change, money move) stay with you 2. Safety monitor can pause or stop the Dot; Activity view shows live tasks 3. Setup path: create in ChatGPT desktop/browser → connect apps → introduce yourself → message or call Risks He Flags (Honest Section) 1. Prompt injection from web pages, emails, documents — content is not supposed to grant permission, but no perfect fix 2. Privacy / training settings on personal plans; Business/Enterprise not used for training by default 3. Still early: cost details, Plus/free access, and long-run reliability still unclear Start Small Advice 1. Connect only the apps you need 2. Set send / buy / delete to ask-me-first in custom rules 3. Review rules often Credit: Javid Shaikh — https://www.youtube.com/watch?v=nhb7mA4sZPw Attached: Dot control + risk checklist.

https://www.youtube.com/watch?v=Q02d6uuqGYM Josh Turner (Construction AI) explains what ChatGPT Dots change for contractor / PM teams — and who should actually switch from a Grok Bot-style setup. Access Reality Check 1. Dots on ChatGPT Pro 100+ and Business Premium; gradual rollout (he is on Pro 100 and had not received it yet at recording) 2. UK / Europe: Pro path limited — Business Premium called out as the way in 3. One Dot for now; more Dots promised later Why Dots Matter In Construction 1. One place to fire messy work — Dot routes instead of staff picking projects, models, and skills 2. Stronger model under the hood (GPT-6 Astra) vs his Grok Bot drawing-test results (~70% vs Claude ~80% in his prior tests) 3. Structure over “just having a go” — construction needs reviewable workflows, not freeform agent guesses 4. Morning brief potential: digests from transcripts, email, registers (variations, subcontractors, assignments) so PMs do not drop commitments 5. Business fit: OpenAI working with Microsoft on Agent 365 control (no date yet) Who Should Switch (His Cut) 1. Stay on Grok Bot if you already have a multi-bot chief-of-staff team working and Dots are not on your account 2. Watch Dots when you need stronger model quality + ChatGPT-native plugins and can wait for multi-Dot / Agent 365 3. Do not treat Dots as a silver bullet — workflows, connected data, and team review still decide results Credit: Josh Turner | Construction AI — https://www.youtube.com/watch?v=Q02d6uuqGYM Attached: Construction Dot go/no-go checklist.