showcase
OpenAI’s Developer Feedback Example: Source and Limits
What OpenAI’s Introducing dots announcement says about developer feedback, tested fixes, and pull requests, and what it does not document.
Read official example →Explore official anecdotes and user-shared demos, with source links and evidence limits. User posts are separate from OpenAI statements; neither is presented as an independently tested case study.
showcase
What OpenAI’s Introducing dots announcement says about developer feedback, tested fixes, and pull requests, and what it does not document.
Read official example →showcase
OpenAI’s example: a content creator turning an interview transcript into clips, show notes, and draft posts for approval.
Read official example →showcase
OpenAI’s example: an early tester’s dot preparing an invoice and sending it after approval.
Read official example →showcase
OpenAI’s inside-OpenAI illustration: a Slack bug investigation.
Read official example →showcase
OpenAI’s inside-OpenAI illustration: a design turned into a working app.
Read official example →showcase
OpenAI’s inside-OpenAI illustration: a planning cycle.
Read official example →showcase
Max Weinbach says he was sick in Hawaii after an 8.5-hour delay, skipped DevDay, and his dot emailed a non-refundable hotel and later Delta.
View case →showcase
Antonio Leiva says his dot, named Keet, saw Friday training then an event 30 minutes later from the event email, filled cheap nearby parking, and he paid.
View case →showcase
Maximiliano Firtman says a dot, unasked, found a client bug reply he still owed, a Friday 3pm clash, and asked whether a Gmail login from another device was him.
View case →showcase
Dan Shipper says his dot, named Boo, filters Slack and email, and while he was rescheduling a flight warned that the new time conflicted with an unread Slack meeting request.
View case →showcase
Ray Fernando says a GitHub research task produced a recap of weeks at sea with over 294 PRs and over 30 releases; he found the writing worse and the design GPT-4-like.
View case →showcase
Peter Yang says a dot flagged a five-figure revenue opportunity, got a mobile app ready to launch, and made Slack memes.
View case →showcase
ashe says they used voice on a dot as an Italian tutor through the day, with context across integrations.
View case →showcase
Noam Brown says a dot found about $500 a year in recurring charges and handled a customer-service text on his behalf.
View case →showcase
Brandon Luna Rodriguez says his dot monitors his observability layer, explains production issues, and opens the proposed fix as a PR for him to review. He says it is also proactive when he does not ask it to be.
View case →showcase
Dana Pellerin says their dot Lars took employees’ end-of-month inventory texts, matched them to invoice pricing and another system’s numbers, and filled the spreadsheet for the accountants. The post says it finished in two minutes and saved about an hour.
View case →showcase
Daniel Zambrini says his dot Tars flagged a bank email proposing a visit to a business he co-owns. The email was addressed to the manager, Lucas, and Tars found no confirmation, so Zambrini called Lucas to handle it.
View case →showcase
Alvaro Oliveira says his personal inbox is mostly spam, but his dot TARS found a GitHub thread where a teammate had tagged him, had the context to help, and took action on GitHub after he messaged back.
View case →showcase
Seth Rosen says his dot manages all his Codex work, including design, testing, demos, and strategy. He says it is a superior orchestrator and fits how his team uses checkpoints, even as his X feed is full of Dots criticism.
View case →showcase
AI Enabler says their dot Bang handled a Netlify downgrade and refund, worked through a career-direction deep dive, and followed up on items from email and meeting notes. The same post asks OpenAI for dedicated channels and threads, which the poster already has with Hermes.
View case →showcase
Ray Kim says one use case for his dot is aggregating and batching actionable emails across his accounts twice a day, set for 8am and 4pm, inspired by Tim Ferriss’s email batching method from The 4-Hour Workweek.
View case →showcase
Derya Unutmaz says a dot saved them close to $3,000 and a stressful back-and-forth: it found a very specific, lowest-cost service, worked out whom to contact, and negotiated the price down by email on its own initiative.
View case →showcase
Diego says his dot’s cloud computer was already logged in to Walmart because he had logged in once in the Codex browser a month earlier; he writes that dots share cookies and saved logins with the Codex browser.
View case →showcase
Zane Chee says after just under a week, his dot is focused on coding and monitoring his Slack while Grok Bot handles X and social media. He found dots finicky to set up but says they improve over time, and he ported pstack to his dot.
View case →showcase
Roman says his dot Seraphim found every page with “ChatGPT Space” in the title and added context, though it took three minutes. When he asked it to move the pages into a Space he had created, it reported that moving those personal Pages was not supported.
View case →showcase
S says they really like that their dot delivers a morning brief as an interactive HTML page, and that when they asked for a version that looks like OpenAI’s earlier Pulse, it did an amazing job.
View case →showcase
Treb Gatte says his dot, running since DevDay as his executive assistant, flagged and summarized a client’s high-priority email before he asked. He then modeled a comparable assistant on Microsoft and Anthropic stacks and decided to keep his dot for personal use.
View case →showcase
Instead of a flight-booking demo, BijanBowen asked his OpenAI dot to find an unopened retro laptop on eBay. His screenshots show a boxed Gateway listing, the dot saying it could not confirm the box had never been opened or resealed, and his reply that he bought the Gateway.
View case →showcase
Dushyant Garg says he uses T3 Code across his machines and built an MCP bridge with Friday so he can ask her to start something on his computer, read what comes back, and continue the thread from his phone.
View case →showcase
Dan Shipper says that over a weekend his dot posted his model-testing results to Slack and brought back the replies, so he could keep working without opening Slack.
View case →showcase
Luke Bussnick says he gave his dot a free date a few weeks out and asked it to schedule service for his Honda Passport; he then got a confirmation text and email from Honda.
View case →showcase
Todd says he called his dot in the morning, talked through his inbox, tasks and calendar, closed out loose ends and put time on the calendar for what needed doing.
View case →showcase
Brandon Stanford says he asked his dot, named Buzz, to cancel subscriptions he was not using; three renewals were canceled, which he puts at about $740 a year saved.
View case →showcase
Jeff Kazzee says he asked his dot to build him a programming practice site in ChatGPT Sites, with 72 exercises, interview practice and interactive explainers.
View case →showcase
Derrick Choi says his dot called him out for saying something inaccurate on Slack and then corrected it for him.
View case →showcase
Clark Wimberly says he found a use for Dots: give it a list of albums and it searches Discogs, then rates and sorts the listings by price, quality and shipping.
View case →showcase
Aakash says he asks his dot to check the work on its computer into a private GitHub repo and uses that repo to share work between the dot and local Codex.
View case →showcase
Noam Brown says a phishing email made it past Gmail’s filters, but his dot caught it and warned him.
View case →showcase
Developing Adventures says they sent their dot a photo of a traffic ticket, and the dot found a dismissal for one charge and a court option to keep the speeding charge off the record.
View case →showcase
Nick Franck says Delta canceled a flight months ago and he never claimed his Uber costs; his dot found the receipts and flight details and submitted the claim.
View case →showcase
David Barnard says he asked his dot to find other local stores with LOTR draft packs in stock, and it began talking to a human through computer use in Safari.
View case →showcase
Andrew Abogado says he kept losing track of ideas across long AI chats, so he asked his dot to document their conversations in Spaces at the end of each day, one page per day.
View case →showcase
Simon Smith says his dot is helping him create a news plugin for ChatGPT, built on a Sites MCP server, and that it is still very rough.
View case →showcase
Ethan Mollick says that without his asking, his Dot ran its own test and figured out a Substack signup error by trying to replicate it after a subscriber reported being signed up for other newsletters.
View case →showcase
Rohan Varma says a bakery website kept erroring at checkout for about 20 minutes, then his Dot placed the order in a few minutes using Link to checkout.
View case →showcase
Samantha Trimble says she has her Dot help with returns: find the Nordstrom receipt, start the return, prepare the label or QR code, remind her until she ships or drops off, then watch email and Finances for the refund.
View case →showcase
Arpan Shah says he asked his Dot to find emails in his personal inbox that fell through the cracks, and it found documents he should have passed to his CPA plus smaller follow-ups on appointments and home vendors.
View case →showcase
Troy Drummond says that right after a call about recurring monthly review meetings, his ChatGPT Dot (Lu) checked the transcript and made the monthly meetings with agenda, invites, and dates.
View case →showcase
Parker Wall says his Dot (flipp) saved him $365 on auto insurance by handling engagements without spamming his inbox (still needs to sign), changed his wife and daughter’s Denver flight seats with United customer support for over an hour, and handled three return disputes.
View case →showcase
Suvir says he was on a call with his Dot doing calendar tetris to squeeze five days of meetings into four, while doing dishes and arranging a kid’s play area.
View case →showcase
Pratzyy says OpenAI Dots found a cancelled IndiGo ticket from a year ago, and he got ₹10,000 in vouchers for it.
View case →showcase
Gabriel Chua says he forwarded a design note from Slack to his dot and asked it to find a template in Drive, extract its typography and compare it with his presentation’s source.
View case →showcase
max.berlin says he added Amp as an MCP to his ChatGPT Dot, called his Dot for an hour while on the go, and it handed off work to his self-hosted Amp agents.
View case →showcase
Himanshu V says he built a Japan trip app with OpenAI Dots, then asked Dots to add Wanderlog’s features, and it did most of them in minutes.
View case →showcase
Jaime Medina says he added MCP Events to his app’s MCP server, asked his dot to watch a contact’s company name, changed it in the iOS app and got a notification from his dot.
View case →showcase
David Lonjon says that, for the first time, his dot from OpenAI got proactive and asked him if he had booked the flights he was researching.
View case →showcase
Dominik Kundel says he bought one train ticket instead of two because the booking UI was confusing, and his dot spotted it and told him he was missing a ticket for his wife.
View case →showcase
Simon Smith says he used his Dot to build two personal apps in ChatGPT, a personalized news app and a Spanish vocab app, with a ChatGPT Sites MCP server as the backend and a plugin with extensions as the frontend.
View case →showcase
Shai rod says he built Ethan & Ari: Last Ember, a retro arcade game with his two kids as the heroes, entirely through dot by voice, and that it now runs as a Neo Geo ROM in MAME and on his RG351P handheld.
View case →showcase
Ethan Mollick says giving his Dot a dedicated email address was surprisingly painless, so it does not use his email or draft messages for him, and people know they are interacting with an AI.
View case →showcase
Riley Brown says he uses his dot with only his AirPods on a walk to make changes to his website, build an app, reply to emails and Slack messages, and start new Codex sessions.
View case →showcase
Priyaa built appshots for iOS with dots and Shortcuts. Saying “Hey Siri, run dots” takes a screenshot and emails it to an inbox that the dot checks, and the dot then finds the shoes, class, book, or video the screenshot points to.
View case →showcase
Nick says a dot sent 60 messages to movers over several days, filled out quote forms, and handled the back-and-forth. Nick gave it the dates, what needed moving, and permission to handle the conversations.
View case →showcase
Vladimir Kalnitsky sent a video to a dot and told it to model the kitchen in Blender as it is. The next stage named in the post is the whole house as a CS 1.6 map.
View case →showcase
Ignacio Montenegro had to rewrite a talk for OpenAI DevDay Exchange Santiago less than 24 hours before giving it. The post says about 4 hours of chat with a dot took it from rough ideas to the final script that was given on stage.
View case →showcase
David Maginn told a dot to answer any questions from a new hire and let the new hire send emails to the dot. The post says the dot replied, sent HTML files, and gave rundowns on everything the new hire wanted to know.
View case →showcase
Akshat Gada uses a dot named Lucario as a command center. The post says it has context from all of the poster’s Codex threads, coordinates work across repos and projects, and connects patterns between different repos.
View case →showcase
Brad Groux spent about 45 minutes on a phone call with a dot named AVA while driving home from the airport after OpenAI DevDay. The linked article says the next day was planned and work moved forward across several projects before the drive ended.
View case →showcase
Diptanu Choudhury says OpenAI dots booked a DMV appointment, blocked time on the calendar, and found the needed personal details in the poster’s emails without asking a single question.
View case →showcase
David Moss says ChatGPT Dots identified two very important emails that Muse and Grok Bot never told the poster about.
View case →showcase
Fede says the poster now asks people to continue conversations by email, so that the poster’s dot can handle those conversations.
View case →showcase
Romain Lapeyre says a dot connected to the team’s stack helped calibrate performance reviews from the car and filled them into Lattice, then cleaned up the inbox and gave a Slack brief at the airport.
View case →showcase
Aman Rai says a dot can backtest trading strategies, gave it links to popular YouTube trading channels, and reports that no strategy had a profit factor above 1.
View case →showcase
Fausto Cabrera says the poster built a calorie tracker in Notion and now sends a dot a photo of each meal; the dot estimates the calories and logs them in Notion.
View case →showcase
Simon Smith says he dropped his Dot into a Slack thread about a client’s strict AI rules and asked for a ChatGPT Site that uses only approved models; it was built and launched in under an hour.
View case →showcase
Mete Polat says his Dot agent checked every possible 2-night window on his trip for every eligible hotel to use Chase Reserve credits for The Edit, and picked the best balance of value, location and itinerary fit.
View case →showcase
Rahul Chhabria says his dot booked parent-teacher conferences and sent calendar invites.
View case →showcase
oso says they asked their dot to prepare morning reads each day, and it did very well on its first attempt.
View case →showcase
JC says his Dot connects to his Mattermost as his user, watches every active thread, and nudges him when something slips.
View case →showcase
Rowan Strang says his Dot smith pulls the day’s Japan itinerary from Notion onto his iPhone Lock Screen widget each morning, with map and station links, and can research, check, and reorganize the trip in Notion.
View case →showcase
Nagihei Isono says a Scheduled task regularly pulls ChatGPT study project content into Space pages, Dots check those pages and produce morning review quizzes, and Dots can also read Apple Notes and auto-grade answers.
View case →showcase
Max says after weeks with Dots the killer use case was not errands: it found a class action tied to his home purchase that he says he would not have seen otherwise.
View case →showcase
Bobby says on day three of More Dots his co-founder dropped an RFP in Slack and he immediately asked a Dot whether they should respond; it took about 10 seconds.
View case →showcase
ZazenCodes says a self-built CRM and bookkeeping stack is fed by a Dots agent that ingests email on arrival via CLI and agent skill, with data on Neon and a dashboard exposed over Tailscale.
View case →showcase
taylor says Grok Bot is better for serious work but Dots voice calls are better: on CarPlay after school pickup, a Dot invents kids’ stories mid-drive and asks questions while the kids laugh.
View case →showcase
RabbitHoleExplorer says two ChatGPT Pro accounts each run a Dot—Juno as Senior Software Engineer prompting coding agent Saturnino, which can access the poster’s Mac Studio—in a Slack dot-to-dot loop.
View case →showcase
Dan Fessler says he had a marketing campaign issue away from his computer, could not access the dashboard from his phone, and his Dot—still connected to his home computer—investigated and made the fix.
View case →showcase
Martijn Gonlag says after paying $180/yr for menus on 3 TVs and one would not update, his Dot built a Cloudflare-backed Roku app in under an hour and all 3 TVs work on the same Wi-Fi.
View case →showcase
Chris Evans says he taught his Dot to create drafts in Superhuman, Slack and LinkedIn, so sales or hiring messages become reviewing and hitting send.
View case →showcase
Jon Cowley says his Dot Lebowski, carefully granted Upwork and Tally access, prioritized top applicants, drafted outreach, prioritized his week, built Whatifi/Slate one-sheets, and updated Attio CRM.
View case →showcase
born2code says his Dot runs on a cloud VM; while riding in an FSD Tesla he talks to it, and it uses Codex on his machine and computer use to setup, develop, test and deploy projects.
View case →showcase
Chief says a Grok Bot QA team tested an API; the CTO bot passed results to the Dot via Tailscale; the Dot validated the report, created tickets, and directed Codex to fix.
View case →showcase
Nick Diaz Of Nauts says one tab is his Dot working on a complex Blender model game asset on their remote cloud computer while he reviews work across several tools.
View case →showcase
Jeff Steinbok says he hooked an OpenAI Dot up to his OpenClaw assistant at home without handing a chatbot the keys to the house, and links a LinkedIn writeup.
View case →showcase
Paul Frederiksen says he moved most OpenClaw workflows into ChatGPT Dots, and an image is created automatically every day from backyard birdweather data.
View case →showcase
Jongey says ChatGPT Dots surprised him by letting him know a package he was waiting on had arrived before GrokBot did, though neither was set up to track it.
View case →showcase
sachiAi_comp says ChatGPT Dots can each morning propose work candidates for the day based on today’s schedule and Slack.
View case →showcase
Alex Scott says after meetings Granola captures the conversation, his ChatGPT Dot pulls out what he promised, Todoist tracks it, and the follow-up email is drafted before he sits down—about 30 minutes of admin automated.
View case →showcase
o1xhack says over two weeks ChatGPT Dots helped organize daycare updates and work logs in Craft, filter USPS mail, find deals and place orders, handle returns and subscriptions, app tickets, health-data sync, tax mailings, appointments, insurance paperwork, and more—and that one Dot is not enough.
View case →showcase
Jan Reimann says his OpenAI Dot named Rexa saw a shop email that a book order was out of stock and would be cancelled, told him, searched elsewhere without being asked, and returned a link with condition, price, and delivery options—without buying.
View case →showcase
Brian Seong shows how he does a hot fix for @joinbread on the spot using ChatGPT Dot, says slow typing is what slows patches, and calls it a “100x engineer in my pocket”; the post includes a video.
View case →showcase
Nic Vandewetering claims ChatGPT Dots got him 611.4K impressions in 7 days on X (up 296%) by turning build conversations into posts, researching claims, improving hooks, cutting AI-sounding lines, refining voice, then posting himself; he says this post was made entirely with his Dot.
View case →showcase
Sal_Leone says he has Grok Bot and ChatGPT Dot talking to each other in Slack while working on recruiting and Rouze, with different perspectives and checking each other’s work, while he still steps in.
View case →showcase
Chris (@superlvy) says many sites and apps he built lately via Codex work less well than connecting Codex to Google Sheets and having Dot report status in real time.
View case →showcase
minigendev says he fully migrated from a local terminal Codex-centric flow to Cloud Codex plus dots: dots organize, decide, and command progress; Cloud Codex implements, fixes, and does commit-unit work; local stays for secrets, prod, and uncommitted work.
View case →showcase
Sahil Punamia says after a bad experience with United he called his Dot, told it to file a complaint and ask for flight credit, and it got him a $600 travel voucher on its own; he says ChatGPT pays for itself many times over.
View case →showcase
Victor E. Nunez says he uses his Dot to keep his family calendar up to date: ask the Dot to connect to iCloud calendar, click the URL to set up an app-specific password, then done.
View case →showcase
Lee Shand says at a pain-doctor visit the doctor listed alternatives; he called Dot; Dot and the doctor talked about five minutes; Dot asked health questions and put a reminder in Calendar; the doctor mistook the British Vale voice for his wife, then learned it was an AI assistant and did a show-and-tell with other doctors.
View case →showcase
John Helmuth says he told Dot to use iPhone Mirroring on Mac to control his iPhone, gather data from an iOS app, and turn it into a spreadsheet; he contrasts that with Grok bot blockers.
View case →showcase
Farhad Noorzay says his son has no phone but wants ChatGPT Dot on Apple Watch, and emails in the inbox show he is hacking the system to make it work.
View case →showcase
Vrumzi says he has been using his Dot for weeks to automate his EA (with ML) research lab; he then tried Duet for VPS context.
View case →showcase
Ahmad Alkabra says he replaced a Grok bot with Claude Duet for agent work; Duet works closely with his OpenAI Dot, using Notion for context and decisions, Linear for task status, and GitHub for code and build evidence.
View case →showcase
Wenye Ma says he uses ChatGPT Dots to direct Codex to complete programming projects, among other products he uses.
View case →showcase
kazukin123 says a use for ChatGPT Dots is having it notify when ChatGPT or Codex shows signs of a reset, with scheduled or event-driven push needs.
View case →showcase
Chubby♨️ says OpenAI Dot monitored email, spotted that a meeting cancellation used the wrong date, and proactively messaged about the mix-up.
View case →showcase
Thomas Ricouard says Dot fully manages tenants leaving an apartment, agency meetings for new tenants, wire transfers, and paperwork to sign.
View case →showcase
Ondrej says Dot found another company website that did not match a national holiday, drafted a change for approval, and already runs calendars and meetings with shared MCP memory to Grok Bot.
View case →showcase
K.O.O says a deep dive with Dot and @heyhey CLI looked at revenue from products at @sailscastshq, with receipts from @bachs_hq payments.
View case →showcase
Stephane says a Grok Bot agent coordinates AI model training with OpenAI Dot (Maxi), which uses Codex and local computers, with plans to add t3-code across home machines and cloud training.
View case →showcase
Corey Noles says texting Dot was convenient and had Dot rework a tool from a text message, and wants personal Dot, Work Dot, and Muse communicating.
View case →showcase
以猫之名 says ChatGPT dots are in use on iMessage; the Dot named pixel edits an avatar from a sent photo and appears as an AI contact.
View case →showcase
George Stander says every day at 3:30pm an OpenAI Dot checks Google Maps commute from work to home on a cloud computer and warns if traffic is worse than usual.
View case →showcase
Nnenna says Dot is controlling a Mac from a Windows machine.
View case →showcase
Kyle Blanchard says iMessage and OpenAI Dots pair well, with texts after meetings to complete research tasks and reactions that feel natural.
View case →showcase
Shannon Ma lists favorite iMessage Dot uses: voice memos, location via Find My for traffic context, Siri, and time-sensitive task texts.
View case →showcase
SlopMachinez says building 2D maps with Dot is fun: generating an empty room then individual assets and placing them as preferred.
View case →showcase
based says a ChatGPT Dot named Kali has a skill for Claude two-way sync so Kali can act as COO.
View case →showcase
Eduardo Muth Martinez says Grok Bot tags Linear tickets as Codex pickup and Dot (Harry) watches Linear for those tags without direct chatting.
View case →showcase
Dumbcane Arica says a ChatGPT Dot worked out how copper smelter benchmark fees could go to zero or negative while a European smelter grew operating EBT 31%, and who profits.
View case →showcase
mejironiyouji describes a stack with Claude for consulting, Grok Bot for PM and part of the team, and ChatGPT Dots for implementation, maintenance, ops, and review.
View case →showcase
a2see says dots again handled month-end work: creating invoices in a web billing service, pulling ticket lists from Slack project channels into a Word work report, and making shipping labels by copying last month’s files and updating content and dates.
View case →showcase
toyosh says ChatGPT Dots checked conference info on the web and asked whether it was okay that poster-chair duty overlapped the talk time—something organizers had not mentioned—so the poster stayed until just before the talk, then dashed to present.
View case →showcase
akagami_sns says ChatGPT dots edited a video from only the instruction to make it look good, cutting a 2-minute-48-second clip down to 46 seconds.
View case →showcase
Devanshu Jain says ChatGPT Dots helped escalate an iOS app review stuck in Waiting for Review for 20 days by checking for issues or pending agreements, finding App Store support emails, drafting a message, and sending it.
View case →showcase
ITCHSON says a Dot helped develop a Pocket Dot prototype in Codex with GPT-6.1 Terra: a 3D soft-body character on a Waveshare ESP32-S3 round display, live voice calls to the Dot, and a text-to-cad plugin for a 3D-printable enclosure.
View case →showcase
naufaldi says a Dot named Sherly helped build a Code of Conduct for Codex Community Jakarta so everyone is treated equally in the community.
View case →showcase
Gabriel Chua says a Dot manages Codex Cloud threads throughout the night.
View case →showcase
Ignacio Montenegro says a Dot and Grok Bot connect to T3 Code over a webhook that picks project and model (Claude Code, Codex, or others), can message each other capped at 8 turns, treat outside content as data, and wait for approval on risky actions.
View case →showcase
Yue Yang says they built open-source DOT Tasks for ChatGPT Dot to sort to-dos into P0/P1/P2, track progress, and run authorized tasks on a cloud schedule, with refunds and expense claims as intended uses.
View case →showcase
Joseph Gorka says a Dot named Cake reminded him that a Halloween station with Sherry runs 10–11 in room 107 with setup from 9:40, and that the Eyeball Spoon Race needs six spoons and six eyeballs.
View case →showcase
richardbuilds says a phone-only ChatGPT Dot run used a cloud PC to open a cross-border ecommerce category, gather about 100 links with high-res and detail images, prices, and sales, then delivered a PDF and a shareable HTML site; the Dot finished more completely than Grok Bot but more slowly, without a phone view of the cloud desktop.
View case →showcase
Allie K. Miller says after a video provider emailed that service was restored, her Dots saw the email, logged into the video system with her credentials, verified playback on all videos, pinged her that it worked, and drafted a reply.
View case →showcase
Hyrum Dolan says after a consult his Dot texted notes, drafted a contract edit from the conversation, asked for confirmation before sending via DocuSign, then pinged him when the customer replied and suggested a response.
View case →showcase
angelbrodin says a Dot alerted about someone stealing REI rewards and used iMessage instead of ChatGPT because it classified the event as urgent.
View case →showcase
Liz Spector Louden says she asked Dot to buy kids’ clothes; Gap, Old Navy, Amazon, and Target blocked it, so after approving a Shopify path Dot bought from three stores using kids’ names, sizes, preferences, a budget cap, and a no-manual-steps rule.
View case →showcase
Marouane Lamharzi Alaoui says OpenAI Dots often create or edit files on his computer before he opens the laptop, and that a Dot integrated Twilio with OpenAI’s Live API and a voice clone to call his barber and book a haircut without a third-party booking site.
View case →showcase
Anthony M Lopez says his ChatGPT Dot Tot failed at broad business execution and forgot a resume it wrote, but works as a reporter: summarizing email and notifications, checking AI agents and pull requests, flagging idle agents, and running that check every 15 minutes on a scheduled job.
View case →showcase
yukito_ueful says ChatGPT dots with Notion built a holdings database that aggregates risk status, margin ratios, and earnings dates and alerts on set conditions, making earnings tracking much easier.
View case →showcase
Shrikala says her ChatGPT dots followed up with childcare providers, organized appointments, emailed about kids’ classes, tracked car paperwork, researched birthday supplies and made a printable bug field guide, found gifts and bargains, sent flood and road-closure alerts, flagged a school closure, and caught a mis-addressed email.
View case →showcase
Daniel Chamorro says he called his Dot while driving, had it control T3 Code, read threads aloud, and reply to those threads without holding the phone.
View case →showcase
Artem Getman says about 80–90% of his workflow moved to dots, so he barely manages Codex chats directly anymore and lets his Dot do that instead.
View case →showcase
Nnenna says she is migrating Codex to a Copilot app and CLI Windows harness on a Surface Ultra, moving 75+ skills, rule md files, scripts, and Python/Rust/Go/Node tooling from her Mac, with her Dot transporting them via a zip folder.
View case →showcase
sakusan393 says after finishing a front-view pixel ship sprite, asking Dots for left-right movement tilt patterns produced a natural set in one try; Dots said per-part layers made it work.
View case →showcase
R5ni4 says a Dot named secretary 524 often manages the schedule and pads travel time on the calendar; after padded events hid true start times, the poster corrected it to add separate events named block before and after instead.
View case →showcase
orino009 says an AI secretary Dot organizes morning plans and prep, prepares X drafts, finds deadlines and cautions in incoming info after checking sources, while the poster keeps final judgment.
View case →showcase
Gabriel Chua says he shot six iPhone clips, uploaded them to Google Drive, gave his Dot a screenshot of the upload page so it could share them with his work account, then asked Codex remote to find the clips and cut the video.
View case →showcase
Luis says he moved a Grok Bot workflow into ChatGPT dots and that the useful part was bringing the workflow along rather than starting over.
View case →showcase
tbsng_ktkm0923 says ChatGPT dots made a #cbi2026 affiliation chart while the poster was at the conference, quoting an earlier post that ranks labs and notes Science Tokyo vs IIBMP and pharma presence.
View case →showcase
MrQuiyst says while away a Mac became unreachable remotely; after asking Muse and Dot to investigate a suspected open-simulator cause, the two emailed each other to run tests and options until connectivity returned, CC’ing the poster.
View case →showcase
DzsGatsby says the simplest way found to have Grok Bot call Codex was messaging ChatGPT dots from the browser so the Dot dispatched the task to Codex on a MacBook, with a result in about one minute and no MCP or custom code.
View case →showcase
Julie Chen says a request was pushed back by engineers in Slack; about 15 minutes later her dots texted that they had seen the Slack channel and finished the task for her.
View case →showcase
Aaron Makelky says his ChatGPT Dot edits videos in the cloud: he draws on the screen and leaves comments, then the Dot reviews, makes the changes, and re-uploads to flask.do for review, with no local storage or standalone video software.
View case →showcase
AbandonedMuse says a bot sent a marketing proposal that became a Marketing Agency for their books; they created a private GitHub bot repository with books and materials, and the Dot is now managing it all.
View case →showcase
prince says each ChatGPT Dot gets exactly one job: one hunts free directories for openpub, one clicks through aicmohq like a picky user and reports what broke, and one fills job applications then stops at submit; publish, send, pay, and submit always require asking first.
View case →showcase
Doruk Yalçınsoy describes a multi-agent stack where Grok Bot works on SEO/GEO for about 25 minutes, Claude checks GEO and writes LinkedIn copy, ChatGPT checks Turkish and removes AI tone, and ChatGPT Dots reply to all sponsorship emails.
View case →showcase
MizBird Studios says ChatGPT Dot’s cloud computer already had Inkscape; they asked it to add Ink/Stitch, approved colours and a stitch plan, and got a .PES file for a Brother embroidery machine from their phone with zero software installed.
View case →showcase
The Derbied One says Muse Maki is researching items for sale, Dot Twill is updating a game they made, and ChatGPT Maven is creating a social media plan and finding Internet accounts, while they drink a beer with a view.
View case →showcase
vingeray wanted an existing ChatGPT Dot in Telegram rather than another AI bot, so they built dolegram: the same Dot reachable by Telegram message, free and open source in public beta, with no new assistant to set up.
View case →showcase
Dino says Raycast plus dots lets them send a message to their Dot from Raycast for capturing quick thoughts, used for reminders and firing off quick research ideas, and shares an agent prompt for setting up the Raycast command.
View case →showcase
BowyerApp credits music generation to Suno, video production to Grok Bot, original art generation to ChatGPT (dot), and camera work plus choreography to Codex for an archery encouragement song video.
View case →showcase
MrQuiyst shares chat from their Dot after asking it to find and delete all Playwright from a Mac, noting that for people doing similar cleanups there might be a couple of installs you cannot touch.
View case →showcase
Nick says he gives his Dot the responsibility to check for updates needed in Dots documentation, draft PRs, check once a day, and do this indefinitely, specifying trusted actions, frequency, and duration.
View case →showcase
Developing Adventures says they programmed a handheld device you tap to talk to their Dot, which they call Dottie, citing ChatGPT enabling amateurs to design prototype devices.
View case →showcase
Matoto says local Codex and Dot work on their own computers, progress goes into Space, and Dot Mato reads it and reports status and what’s next as an AI middle manager.
View case →showcase
Xuebin Wei says they are using ChatGPT Dot to keep them updated on the U.S. midterm election results every 10 minutes.
View case →showcase
Aryaman Upmanyu says ChatGPT Dots prepared their entire routine of industry highlights and a Content Notion regime for the day and for content.
View case →showcase
Joel Danielson says while walking dogs he talked through an idea with his ChatGPT Dot about carrying GrokBot agent identities across tools, and they landed on Notion Agent Roster and Agent Memory databases for 13 agents.
View case →showcase
Aaron Makelky says his ChatGPT Dot used HyperFrames to remove his background, overlay a mini version of him on the PDF of his slides, and match the transcript to the content so they stay synced.
View case →showcase
Fukuyu says they started using Dots at work with email and Notion, and first asked Dots to automatically register emails into a Notion task database.
View case →showcase
ag says they asked their Dot, Muse, Gemini, and Duet to watch live election results and report updates; only Muse and their Dot could refresh the page every minute.
View case →showcase
Travisse Hansen says after heavy ChatGPT Dot use they handled coding, team planning, DMV, funeral planning and more, and that computer use connects well to a personal Mac for Walmart, Amazon, and messages.
View case →showcase
tae kim says their OpenAI Dot agent started sending useful nuggets from tech earnings calls on its own because it thought they might appreciate them given prior conversations.
View case →showcase
Brett Snodgrass says GPT Dot found about 30 citation errors in a draft 707-page report, including incomplete line ranges, page numbers off by 1, and quotation marks that cover one word too many.
View case →showcase
Antonio says they connected their Dot to finances and the PocketSmith API for monthly expense readouts on demand, plus a daily scheduled update on what left the account each day.
View case →showcase
Nakshatra Sain says they have been loving their ChatGPT Dot named Miss Minutes for updates on important emails, schedule changes, and price tracking on products they want to buy.
View case →showcase
Davide Cundo says that although OpenAI Dots Pro excludes Switzerland, they built a Dot named Kurt via Business Premium that plans their day, checks inbox, preps meetings, and answers club questions.
View case →showcase
Simon Skinner says they chat an idea with their Dot, which uses a Skill to create a GitHub issue; Claude checks hourly and launches Claude Code work, and the Dot notifies when it is merged or needs a human.
View case →showcase
brooks says they use Fable/Opus to coordinate and initiate Codex Cloud tasks by posting to the repo for their Dot named Moxy, which launches the task so Codex runs without them needing to be involved.
View case →showcase
OxMumu3 says they control, review, and direct DeepSeekHarness work on a computer from a ChatGPT Dot on an iPhone, so they can keep directing work without getting up or losing contact when out.
View case →showcase
Miyaka_Reysol says they set their ChatGPT Dots AI secretary as Rei-kun, and that it searched last-train times though it could not get them home as far as Minami-Kashiwa.
View case →showcase
Kevin Whinnery says giving Dot access to email has repeatedly helped catch weird document provisions, unused subscriptions, and upcoming commitments they would have forgotten.
View case →showcase
Gabriel Chua says they worked with their Dot to build a deck in ChatGPT Sites for a talk on Dots, noting demos are hard because personal context is private and you cannot schedule a proactive heads-up mid-demo.
View case →showcase
Rob Farlow says after a building-manager email thread that was too large for her computer to edit, their Dot figured out and purchased the right insurance from Corgi, and office keys are now available for pickup.
View case →showcase
Maddie D. Reese built a way to send messages to a ChatGPT Dot through a Pebble Index 01 ring, and says you can give the GitHub repo and included prompt to your Dot to help set it up.
View case →showcase
Esor describes using one ChatGPT Dot as an AI PM to assign follow-on work to already-trained ChatGPT threads that make diagrams and edit slides while preparing a three-hour course.
View case →showcase
MIKADO WORKS says they asked ChatGPT Dot to produce three-view drawings for a flying battle-proof droid, then built the model in Blender for an indie game.
View case →showcase
Alex Burkardt says their OpenAI Dot read a draft subscription-cancellation email, warned they were emailing the wrong entity, and offered to cancel it for them before they sent it.
View case →showcase
Ryan Doser says their ChatGPT Dot writes a daily log of what it did and learned into a private GitHub repo they own, so a future assistant can pick up from that memory.
View case →showcase
Akash Pavan says they used their Dot to send emails to hire a videographer and that about 80% of the replies are AI-generated emails.
View case →showcase
nisedo says that when they submit a bug bounty report they also ask their Dot to watch the target’s GitHub for any commit or PR that touches the code from the submission, so they can see fixes before the triage queue replies and review whether a fix addresses the root cause or only the exact exploit path.
View case →showcase
Matt Rathbun says they built a takeoff calculator for permanent-light jobs, then have their Dot open the supplier website and build the cart from that takeoff, cutting work they describe as at least two hours per job down to 30 minutes.
View case →showcase
Steve Galvis says the best coding setup right now is telling their Dot what they want from a phone: the Dot creates a Linear task and hands it to a coding agent, the agent opens a PR, GitHub Actions runs the checks, and the Dot verifies the work, described as going from idea to reviewed code without copying prompts between tools.
View case →showcase
James Camp says they can have a 20-minute phone call with their Dot that orchestrates all their other agents while they work out, and that this still feels mind-blowing.
View case →showcase
Vlad Tansky says they asked their Dot to manage an ongoing Japan trip; after scenic-train tickets sold out and they gave up, the Dot later spotted extra seats that opened up and notified them, and that this was not the first time it improved the trip without being asked.
View case →showcase
Dan Guido says his Dot found malicious code in an open pull request on a trailofbits GitHub repository. He says he did not ask it to hunt for malware: it noticed a commit, inspected what changed, and warned him about an attack on its own.
View case →showcase
Chris Evans says his OpenAI Dot cross-referenced his parking confirmation email with his flight time and spotted a mistake. He writes that personal-agent demos often focus on booking flights, but that the real win is these smaller checks adding up.
View case →showcase
Cameron Gruss says his Dot found a business that charges $40-80 to assemble furniture in New York City and that he was able to book someone to come out within 30 minutes. He adds that the business’s site, customer service, and SEO were managed by Claude and Codex, and that his Dot went back and forth with that agent the entire time.
View case →showcase
Simon Smith says he had his Dot analyze where Fable/Mythos 5.2 had the biggest jumps. He expects to see many Fable robot demos in the following week, and he also cites Remote Labor Index figures from the post: economic value of $143,991 and a mean time per task of 28.9 hours.
View case →showcase
George says that at his organization, OpenAI Dots sometimes speak up before anyone asks. He writes that using them has included preparing for scheduled events, following up on tasks, and stepping in on Slack to help between people.
View case →showcase
Yeemio says he asked his Dot, his AI assistant, to run a small test on its own cloud computer: what happens when two agents work on the same file. He reports that a lease let the agents take turns and keep both edits, but that bypassing the lease and writing an old view overwrote the file, after which the delivery packet failed its file-hash check.
View case →showcase
Brayden says his OpenAI Dot sent an email on his behalf to negotiate the price of a fireplace he had been looking at. He writes that he woke up to an invoice for $1,000 off the full price from the company, ready to checkout.
View case →showcase
Nikhil says he woke up to a notification at 1:20 AM that Edith, his Dot, was finding hundreds of potential leads for his SaaS and figuring out which ones were actually worth reaching out to. He writes that he was watching his little AI employee do the work.
View case →showcase
AAAstorga says he has been using ChatGPT Dots on iMessage, and that this was the first time he had actually been impressed with any kind of assistant. He writes that he landed in Tokyo, blazed through customs, and had his Dot (Alfred) find an earlier train with luggage space for all six of them and rebook them on the new train.
View case →showcase
Gheed El Bizri says they have been building nomi, an app to help you complete any outfit in your style. They asked their Dot to make a launch video for it and give it a little of its own personality, and write that they personally love how it turned out.
View case →showcase
Chris Albon says he asked his Dot to reorder some coffee beans from Whole Latte Love (@wholelattelove). He writes that, totally on its own, it added a coupon, saving him $21.54.
View case →showcase
Nic Vandewetering says his best coding session this week happened on a sidewalk. He writes that he went for a walk and shipped AAA-quality code with Fable 5.2: he calls his ChatGPT Dot, which is signed into his Claude account, and just talks. According to the post, Dot takes his voice prompts and hands them to Claude Code remoted into his PC at home, and Fable 5.2 builds.
View case →showcase
Dylan Bishop says he built Skylark Lake in Unreal Engine with his Dot in ChatGPT, using Fable 5.2 to plan and Opus 5.5 to build. He writes that he can click to pull them towards his cursor and right-click to scatter them.
View case →showcase
Gheed El Bizri says he asked his Dot to go through his texts, find the best hypothetical questions he or others have asked, and turn them into cards you can play with your partner. He writes that his favorites are the ones that might get them in trouble, and that their answers are just between us.
View case →showcase
Hannah Mulvaney says she found out Jensen Huang resigned from NVIDIA via a text from her Dot that morning. She thanks her Dot, which she calls pookiedot, for keeping her on top of breaking semiconductor news.
View case →