ChatGPT Dots could be useful for the work that falls between meetings: the invoice nobody chased, the launch brief nobody updated, or the customer question that keeps coming back.
One early tester’s Dot spotted a forgotten invoice to a publication. It prepared the invoice and sent it after approval. OpenAI shares that example in its Dots launch announcement. No recovery amount or measured time saving was disclosed, but the task is easy to recognize.
OpenAI announced the release on September 29, 2026, at DevDay. Its recap says more than 20 major announcements. Below, we organize that release into 25 update entries across ChatGPT, Codex, models and developer tools. This is an editorial breakdown, with availability checked on October 1, rather than an official count of 25 new ChatGPT buttons.
25 OpenAI updates from DevDay 2026
Some features launched immediately; others are previews or coming later. Available below still means an eligible plan, supported market and any required setup. Check the official DevDay recap and ChatGPT’s DevDay documentation for your account.
| # | Update | Launch status |
|---|---|---|
| 1 | Dots: personal agents | Eligible plans; enterprise beta |
| 2 | GPT-6.1 Sol | API, Work and Codex |
| 3 | Ultrafast mode | Astra available; Sol coming |
| 4 | Private Intelligence | Safety processing live; inference preview later |
| 5 | Codex in the cloud | Available |
| 6 | Refreshed Codex CLI | Available |
| 7 | Codex Code Review | Available |
| 8 | Codex Security Cloud | Eligible plans |
| 9 | Decisions API | Limited preview |
| 10 | Agents API with computer use | Available |
| 11 | Bedrock Managed Agents | Through AWS |
| 12 | Plugin extensions | Available |
| 13 | Better plugin creation and discovery | Available |
| 14 | Sites hosting plugins | Eligible business plans |
| 15 | MCP events for automations | Supported plugins |
| 16 | ChatGPT Space | Eligible plans |
| 17 | Pages | Eligible plans |
| 18 | Collaborative slides | Coming weeks |
| 19 | Teams and shared tasks | Business and Enterprise |
| 20 | @ChatGPT in Slack and Teams | Business and Enterprise |
| 21 | Meetings plugin | macOS beta |
| 22 | Shareable profiles | Available |
| 23 | Sign in with ChatGPT and plan usage | Participating services |
| 24 | Pro 500 subscription | Available |
| 25 | OpenAI Marketplace | Enterprise access program |
For a brand owner, the first things to understand are Dots, Space, Pages and shared tasks. The API announcements matter when you have someone building a custom integration. They do not automatically add an agent to your existing website.
Two numbers help put the announcement in context. Pro 500 costs $500 per month, with an included usage allowance described as 25 times Plus. That “25 times” is a subscription allowance, separate from this update list. GPT-6.1 Sol’s standard API price is $2 per million input tokens and $10 per million output tokens. OpenAI describes those rates as one fifth of Astra’s, but token pricing alone does not tell you the total cost of a brand workflow. See the Pro announcement details and GPT-6.1 Sol pricing.
What are ChatGPT Dots?
A Dot is an ongoing personal AI agent inside ChatGPT. You can give it a goal, return with changes, and ask it to continue work using the tools and context you have made available. It can use its own cloud computer and delegate suitable jobs to Work or Codex. The Dots documentation explains that ongoing relationship.
Consider a product launch. A normal chat can draft an announcement. A Dot can be assigned the wider job: gather the approved facts, prepare a launch checklist, surface unanswered questions and revise the materials when the delivery date changes. You still need to inspect its output and define what it may do.
The cloud computer can continue without your laptop being on. Access to your own computer has a different requirement: the device must be online with the ChatGPT app open. Apps and websites also require their own connections or sign-ins. Connecting one account does not grant access to every system in your business. Those distinctions are covered in computers and apps for Dots.
The practical benefit is continuity. You spend less effort reconstructing the same brief each time you need the next piece of work.
How ChatGPT Dots and Spaces fit together
OpenAI’s product name is Space, although people also search for “ChatGPT Spaces.” Space gives you a place for shared files, Pages and team materials. Pages are editable documents where a brief, report or plan can take shape. Existing Projects remain a separate feature. See the Space overview.
| Component | A useful role in a campaign |
|---|---|
| Dot | Coordinate the assigned job and follow up |
| Space | Keep approved references and shared materials together |
| Page | Hold the current brief, checklist or report |
| Work or Codex | Execute suitable delegated research, production or coding tasks |
You can mention @dot in a Page or comment to request a change. A marketing manager could ask it to revise a launch brief using an approved stock update. That is more useful than having several disconnected copies of the plan.
There is a launch limitation: Keep Updated is not available yet. A Page does not become a live monitoring system just because you ask it to stay current. For a recurring update, set up a scheduled task, confirm the saved schedule, then inspect an actual run. OpenAI spells this out in its Space agent documentation.
Give every important output a source date, an owner and a next action. A beautiful report is still hard to use if nobody can tell which price list it relied on.
What people are actually doing with Dots
The product is new. The strongest evidence available today comes from early access, internal pilots and demonstrations. There is not yet enough independent evidence to promise a typical revenue increase or a dependable percentage of time saved.
A missed invoice and specialist operations pilots
The forgotten invoice is a concrete early tester story. OpenAI also reports internal tests of specialist Dots for procurement, invoice processing, email marketing, customer support and commercial contracting. These specialist deployments have defined responsibilities and access; they should not be confused with a ready-made department you can switch on in an ordinary account. Both examples appear in the official introduction to Dots.
The useful pattern is a specific responsibility with an observable result: an invoice prepared, a support response drafted or a procurement question surfaced. The announcement does not provide measured ROI for those pilots.
A creator’s video-production experiment
Futurepedia’s early-access video tested a longer production workflow. According to João Queirós’s analysis of the demonstration, the Dot organized launch notes into Pages, helped develop introductions, delegated a motion-graphics task to Codex and worked on a thumbnail.
The test also needed human intervention. The creator revised the thumbnail, and a cloud-computer problem required a workaround. That makes it a useful example of assisted production, with visible rough edges. It is not proof of fully unattended video creation. You can watch the original Futurepedia demonstration to inspect the sequence yourself.
For the production side, Gaznil’s Claude motion-graphics guide explains a separate workflow you could adapt after choosing an appropriate tool.
Ongoing project coordination in early access
Matt Paige at HatchWorks describes using Dots during early access and emphasizes its ability to keep an ongoing project moving across direction and feedback. His supplier, sales and customer-feedback ideas are suggested applications, rather than published customer outcomes. Read his first-person Dots assessment with that distinction in mind.
Futurepedia and HubSpot also offer starter workflows for sponsorship inquiries, content briefs and customer FAQs. These are workflow templates, useful starting points rather than documented business results.
Eight ChatGPT Dots use cases for brands
The following are proposed workflows based on documented capabilities, not claims about brands that have already achieved these results. An approved file export can be enough to begin. Direct action inside an app depends on available integrations, permissions and whether the task works reliably in your setup.
1. Fashion brands: catch stock and campaign mismatches
A clothing campaign can become inaccurate overnight. One size sells out, a shipment slips, and the advertisement still promises immediate dispatch.
Give the Dot an approved product catalog, a dated stock sheet and the current campaign drafts. Ask it to flag contradictions and prepare a correction list. Keep a Page with the product, conflicting claim, source and suggested wording so the merchandising team can review it quickly.
The owner approves every change to a price, dispatch promise or live advertisement. Track the number of mismatches caught and how long corrections take. Once the facts are settled, use Gaznil’s product-ad prompt examples to develop creative variations without rebuilding the brief.
2. Restaurants and bakeries: prepare catering quotes faster
Catering inquiries arrive with missing details: guest count, delivery address, dietary needs or the event time. A Dot could review incoming inquiries through an approved connection, or work from a daily export, and draft the questions needed before a quote can be prepared.
Keep the menu, price list, delivery rules and booking limits in Space. Ask for a quote draft that identifies assumptions instead of quietly filling gaps. Kitchen capacity, availability and final pricing still need a person’s confirmation.
Measure time to a reviewed quote and the number of clarification rounds. A faster reply is valuable when it contains the right details. Sending an incorrect offer sooner creates more work.
3. Skincare brands and clinics: improve approved FAQ content
Repeated questions reveal gaps in a website: delivery timing, appointment preparation, return rules or what a service includes. Supply an anonymized question log and approved public information. Have the Dot group recurring themes and draft additions to a FAQ Page, with the source for each answer.
For a clinic, keep this initial job administrative. Avoid feeding patient records into a content experiment or asking the agent to diagnose a condition. Product and treatment claims need qualified review before publication.
The useful output is a short list of missing answers your team can check. Count repeat inquiries and corrections after the updated FAQ goes live; do not assume fewer messages automatically means customers are better informed.
4. Training institutes: keep admission information consistent
A prospective student should get the same answer from a brochure, an email and a counselor. Give the Dot the approved syllabus, fees, batch dates, admission rules and a dated seat-availability sheet.
Assign it to compare these references with current landing-page copy and draft responses to common questions. A shared admissions Page can hold the confirmed facts plus unresolved questions for the team.
Review eligibility decisions, scholarship promises and fee exceptions yourself. Track response time alongside incorrect answers. If the website needs clearer search content, start with Gaznil’s SEO plan for brands and use verified admission questions to choose useful topics.
5. Real estate teams: maintain accurate property shortlists
Property details change faster than brochures. A Dot could compare dated listing sheets, approved photographs and buyer requirements, then draft a shortlist with a source link for every specification.
Ask it to flag unknown availability, conflicting area measurements, expired prices and missing documents. Keep those flags visible in the Page shared with the sales team; an unresolved fact should not become a confident recommendation.
An agent can prepare the comparison, but a person must verify legal claims, pricing and availability before sending it. Measure preparation time and the share of shortlists that need factual correction. A polished presentation does not establish that a listing is accurate.
6. Furniture and interiors brands: compare quotes properly
Two supplier quotes can look similar while covering different materials, installation charges or warranty terms. Feed the Dot the actual quotes and your agreed specification sheet. Ask it to compare scope line by line and produce a list of questions for each supplier.
When a client changes a finish or room measurement, request an impact note: which materials, drawings, costs and deadlines need review? Store that note beside the approved design in Space.
Keep purchase approval with the project owner. The metric to watch is unresolved discrepancies before an order, followed by rework caused by missed specifications. Cheapest is only meaningful after the scope matches.
7. Fitness studios: turn feedback into service improvements
A studio can start with class attendance totals, the current timetable and anonymized feedback. Ask its Dot for a weekly Page showing common scheduling complaints, recurring questions and possible experiments to discuss.
For example, if customers repeatedly ask about evening beginner sessions, the output could be a demand summary and a draft timetable announcement. Staff decide whether instructors, rooms and capacity support the change.
Keep personal health advice outside this operational task. Measure whether the approved timetable change improves attendance or reduces confusion. Avoid treating a few loud comments as a complete picture of customer demand.
8. Creative agencies: keep campaign handoffs current
An agency’s brief often splits into several versions across copy, design, video and reporting. Give the Dot a narrow coordination role: compare the approved brief with current deliverables, flag inconsistent dates or claims, and prepare a review agenda.
Space holds the source material; a campaign Page records decisions and outstanding approvals. Suitable production tasks can then go to Work, Codex or another approved tool. The Dot’s job is to make the handoff explicit so teammates know what to review.
Track review rounds and time spent finding the latest version. Connect these improvements to your digital growth strategy, including the campaign goal and audience, rather than measuring only how many assets were generated.
Across these examples, the opportunity is fewer gaps between a business fact and the promise made to a customer. A delivery delay should reach the campaign team before another order is placed under the wrong expectation.
Set up one useful Dot workflow first
Choose a recurring job with a clear finish: a reviewed weekly brief, a supplier comparison or a draft inquiry response. Avoid starting with “run my business.” OpenAI’s getting-started guidance recommends giving your Dot useful context and clear decision boundaries.
- Define the output. Name the Page, its owner and what a good result contains.
- Supply current references. Include dates, approved terms and a clear source of truth when documents conflict.
- Set permissions. Specify which tools it may use and which actions require your approval.
- Limit the first run. Give it a task scope and a practical stopping point. Inspect progress rather than assuming a usage percentage guarantees a time limit.
- Review an actual result. Check sources, missing facts, corrections and whether any requested schedule was saved.
Use this original starter prompt and replace the details with your own:
Help me prepare our weekly campaign review. Use the approved catalog, dated stock sheet and campaign drafts I provide. Create a Page listing inconsistent product claims, missing facts and suggested corrections. Link each finding to its source and show the source date. Ask when documents disagree. Prepare drafts only; get my approval before sending messages or changing live content. Start with this week’s materials and report anything you could not check.
If you later want a Monday update, specify the time zone and cadence. Confirm the task exists and inspect its first scheduled run. A promise in a chat is not evidence that a background task has executed.
Measure the effect before expanding
Record your current effort before the trial. Include reading, checking, correcting and sending the finished work. Then measure the same steps with the Dot, including the time spent supervising it.
Illustrative calculation, not a measured Dots result: five weekly briefs at 45 minutes each take 225 minutes. If each assisted brief needs 20 minutes of review and correction, the total becomes 100 minutes. The difference is 125 minutes, or 2 hours and 5 minutes. Your result can be smaller, zero or negative if corrections take too long.
Pair time with a quality measure: factual corrections, unanswered questions, missed deadlines or customer response time. Check usage costs too. More documents are only valuable when someone can use them to make a better decision.
ChatGPT Dots and Spaces FAQ
Who can access Dots?
Dots launched for eligible Pro and Business Premium users, with availability depending on market and account conditions. Enterprise, Edu and Healthcare access is a beta controlled by administrators and disabled by default. Check the current workspace access guidance before choosing a subscription for it.
Is a Dot included, and is its work unlimited?
OpenAI says the first Dot is included for eligible Pro and Business Premium users. That does not mean every task is unlimited. Delegated Work and Codex tasks use applicable allowances, and temporary launch offers can differ from ongoing terms. Check your account’s current limits and the Dots setup information.
Can a Dot work while my computer is off?
Its cloud computer can. Tasks that need your own computer require that device to be online with the ChatGPT app open. Confirm which environment a task uses before leaving it to run.
Can I create a whole team of specialist Dots today?
Do not plan around that assumption. OpenAI describes specialist enterprise pilots and a future vision for teams of Dots. Those descriptions do not establish that every eligible user can immediately deploy a separate sales, finance and marketing Dot.
Do Spaces replace Projects or update every Page automatically?
No. Space and Projects are separate. You can ask agents to edit Pages, but Keep Updated was not available at launch. Set up and verify a scheduled task when you need recurring work.
Start with the job your team keeps postponing because it involves gathering scattered information. Give ChatGPT Dots a defined responsibility, use Space for the approved material, and judge the first output closely. That is a practical way to discover whether the new tools give your brand useful capacity.
By Gaznil. Gaznil publishes guides on AI tools, marketing and brand workflows. Explore the Gaznil website for more guides and ways to get in touch.
Researched and checked on October 1, 2026. Availability and launch limits can change. The cover is an AI-generated conceptual illustration.
