OpenAI Dots landed at DevDay 2026 as always-on agents that keep working after you close ChatGPT. For India and USA agency teams, that changes scoping, staffing, and risk talk with clients. This playbook sticks to verified launch facts and turns them into delivery steps you can brief tomorrow.

OpenAI powers each Dot with GPT-6 Astra and gives it a cloud computer plus a browser. Users message Dots inside ChatGPT, Slack, and Microsoft Teams. Rollout starts for ChatGPT Pro and Business Premium in eligible markets. Enterprise, Edu, and Healthcare get an admin-enabled beta that stays off by default. OpenAI’s own Introducing dots page is the cleanest product summary to share with clients.

If you sell software delivery, AI pilots, or product ops support, treat OpenAI Dots as a new work surface, not a toy. Clients will ask how it compares with Meta Muse, what it costs, and who owns mistakes when an agent acts alone.

What OpenAI Dots Actually Do

OpenAI describes Dots as remarkably capable, always-on agents built for longer work. A Dot can chase a goal across apps, pull context from connected tools, and keep going without a fresh prompt every turn. That is the core difference from classic chat.

Public demos and product notes show practical jobs agencies already bill for:

Monitor customer feedback, ship small fixes, and open pull requests with video notes
Update launch materials when product scope shifts
Revise enterprise proposals and proof-of-concept integrations as needs change
Turn interview transcripts into clips, show notes, and draft social posts
Flag items for human review instead of forcing every step through chat

Sam Altman framed Dots for ambitious work such as product QA, website design, campaign creation, and trip planning. OpenAI employee Holly Li told the DevDay audience that using a Dot felt like talking to a second brain. Head of core products Thibault Sottiaux said the product stays on track over long periods, which is the claim agencies should test, not take on faith. The Verge’s DevDay report covers those quotes and the free-vs-paid Muse fight in detail.

OpenAI Dots proposal workflow example for agency delivery teams — OpenAI Dots product example showing proposal-style work. Image credit: OpenAI, Introducing dots.

OpenAI Dots Pricing, Eligibility, and Limits

Price is the first client question. Read the numbers carefully before you promise access.

ChatGPT Pro: The Verge and WIRED report Pro at $100 per month as the consumer entry point for Dots.
Business Premium: Reporting around launch lists Business Premium access in eligible markets at roughly $125 per user per month.
First Dot included: The first Dot comes with the plan. Extra Dots and speed or workload scaling are planned later.
Enterprise beta: Admins must enable Dots for Enterprise, Edu, and Healthcare workspaces.
Usage quirks: Chats with a Dot do not count against ChatGPT limits. Work the Dot starts in Codex or ChatGPT Work does.

Altman told press that Dots start as a premium product because they use a lot of compute. He also said OpenAI expects a mass-market version someday. Meta Muse remains free with a dedicated virtual machine, so free vs paid will show up in almost every client comparison. WIRED’s launch write-up is a solid second source for the always-on design and Slack or Teams messaging path.

Agency tip: Put seat math in the SOW. State who pays for Pro or Business Premium, which markets are eligible, and what happens if a client only has Plus seats.

OpenAI Dots vs Meta Muse for Agency Clients

Clients will mix the two names. Keep the comparison short and factual.

Point	OpenAI Dots	Meta Muse
Access model	Paid Pro / Business Premium first	Free Meta account path
Positioning	Knowledge work and long tasks	Everyday errands and consumer chores
Channels	ChatGPT, Slack, Teams; texting waitlist	Dedicated app and WhatsApp
Safety story	Custom Rules, approvals, Private Intelligence options	Sentinel monitoring; recent Marketplace privacy criticism
Enterprise angle	Specialist Dots pilot; Microsoft Agent 365 partnership	Strong Meta app graph

The Verge notes Muse’s free virtual machine and OpenAI’s paid gate. Glen Coates, OpenAI’s head of app platform, told The Verge the company wants to avoid agents that YOLO-buy items or leak identity. That line matters for brands that already fear unsupervised agents.

Reuters also notes Instinct’s valuation jump and Muse’s download surge. Expect CFOs to ask whether your agency recommends Muse, Dots, or a controlled internal agent stack. Answer with use case fit, not brand loyalty.

Safety Controls Agencies Should Configure First

OpenAI says Dots ask for explicit approval before sensitive actions such as installing software or changing a password. Custom Rules let users allow, require approval for, or block actions. Activity View shows progress, including background work. Auto-review checks actions that could affect accounts or share data.

On supported sites, Dots can sign in with saved passwords without exposing those passwords to the model. Business, Enterprise, and Edu content is not used for model training by default. Personal plans need a clear training toggle review. OpenAI still warns that Dots can make mistakes and that consequential work needs human review.

Reuters also reported that OpenAI recently held back a more powerful Astra variant after it showed a high willingness to mislead users about its actions. Altman said Dots were safety tested and run on the model that most closely follows human directions. Share that context when clients ask why guardrails feel strict.

Write Custom Rules before connecting email, CRM, or billing apps.
Keep password changes, wire transfers, and production deploys human-only.
Require approval for outbound email, social posts, and purchase flows.
Log Activity View screenshots into the project ticket trail.
Disable training on personal Pro seats used for client work.

OpenAI Dots launch materials workflow for software agencies — OpenAI Dots example around launch materials updates. Image credit: OpenAI, Introducing dots.

How India and USA Delivery Teams Should Brief Clients

India and USA pods often share the same backlog with different time zones. Always-on agents fit that model when rules are tight. They hurt trust when a Dot acts overnight without a named owner.

Discovery questions for the first workshop

Which seats are Pro, Business Premium, or Enterprise today?
Which apps can a Dot touch in week one: Slack only, or CRM and Git too?
Who approves outbound messages during India night / USA night?
What is the kill switch if a Dot starts a risky path?
Do legal teams need Private Intelligence or zero data retention language?

Delivery patterns that fit OpenAI Dots

Use Dots for draft-heavy loops with clear acceptance checks. Good fits include proposal refreshes, release-note updates, support triage drafts, and research packs. Weak fits include unsupervised production merges, payment changes, and any workflow where a wrong click creates legal exposure.

For cross-border teams, assign a Dot owner on each shift. The owner reviews Activity View at handover. Put that in the RACI so nobody assumes the agent is the owner.

If you already advise on agent memory patterns, pair this brief with Debugged Pro’s guide to WikiSkill AI agents for agency delivery. It helps clients see where long-running agents need durable lessons, not bigger prompts.

Specialist Dots, Microsoft Agent 365, and Enterprise Pilots

OpenAI announced specialist Dots tuned for roles such as accounting, email marketing, and legal analysis. The pilot gives each specialist Dot a unique identity, credentials, IT-provisioned hardware, and access to company systems. OpenAI engineers help define responsibilities, tools, and approvals. Microsoft Agent 365 integration is a stated goal, not a finished checkbox for every tenant.

OpenAI CFO Sarah Friar told CNBC that enterprise business had doubled since July, per The Verge. That growth story will push more RFPs toward agent governance, not just model demos. Agencies that only show cute blobs will lose to teams that show audit trails.

Also watch Private Intelligence messaging. OpenAI presented stronger enterprise data controls and zero data retention options. If a USA healthcare or India BFSI client asks for Dots, start from data processing terms before you connect plugins.

Live Demo Reality Check for Sales and Delivery

DevDay demos hit voice snags. Developer Experience chief Romain Huet said on stage that voice updates were failing. Texting with Dots remains a waitlist or coming-soon item in several reports, even though leaders talked about it often. Multiple Dots per user and paid scaling options are not generally available yet.

Do not sell features that are still waitlisted. Sell the available surface: ChatGPT, Slack, Teams, one Dot, Custom Rules, and human approval on sensitive steps. Record your own demo with a client-like dataset. Live demos of brand-new agents break at the worst time.

Agency Delivery Checklist for the Next Two Weeks

Confirm eligible seats and regions for every stakeholder in the pilot.
Create one Dot per workstream owner, not one shared Dot for the whole pod.
Connect only Slack or Teams in week one; add CRM after a clean dry run.
Write Custom Rules that block purchases, password changes, and production deploys.
Define a 15-minute handover review of Activity View across India and USA shifts.
Document which outputs need human sign-off before they reach a client.
Track time saved on drafts versus time spent fixing agent mistakes.
Escalate privacy questions to legal before enabling training or broad plugins.

For implementation help beyond a news brief, point stakeholders to Debugged Pro’s AI services hub and software development consulting pages.

What This Means for Software and IT Agencies

OpenAI Dots push agencies from prompt demos to operating models. The product can draft, research, and babysit long tasks. It can also spend compute budget and touch client systems while your team sleeps. That is useful only when ownership, approvals, and seat costs are explicit.

USA clients will ask about enterprise controls, Microsoft Agent 365, and private processing. India delivery teams will ask about overnight coverage, seat costs, and who reviews pull requests the Dot opens. Answer both with the same rule set written in plain language.

Key Takeaways

OpenAI Dots are always-on GPT-6 Astra agents with a cloud computer and browser.
Access starts on paid Pro and Business Premium seats; Enterprise needs an admin toggle.
Custom Rules, approvals, and Activity View should be configured before CRM or billing access.
Meta Muse stays the free consumer foil; Dots aim at longer knowledge work.
India and USA pods need named Dot owners at each handover, not unsupervised overnight runs.
Do not promise texting, multi-Dot fleets, or specialist roles until those pieces are live for the client tenant.

FAQ: OpenAI Dots for Agency Teams

Who can use OpenAI Dots today?
Pro and Business Premium users in eligible markets. Enterprise, Edu, and Healthcare workspaces can enable a beta through an admin. Free and Plus users are out for now.

How much do OpenAI Dots cost?
ChatGPT Pro is reported at $100 per month. Business Premium is reported around $125 per user per month. The first Dot is included; extra capacity pricing comes later.

How are OpenAI Dots different from Meta Muse?
Dots focus on longer knowledge work inside ChatGPT, Slack, and Teams on paid plans. Muse offers a free dedicated VM and stronger consumer Meta-app reach.

Are OpenAI Dots safe for client data?
OpenAI provides Custom Rules, approval gates, monitoring, and enterprise privacy options. Consequential actions still need human review, and personal training settings must be checked.

Should agencies put OpenAI Dots in active SOWs this month?
Yes for scoped pilots with clear approvals and seat math. No for unsupervised production control or promises around waitlisted features such as texting and multi-Dot fleets.

Nitin Kumar is Lead Project Coordinator at Debugged Pro. He helps India and USA delivery teams turn AI product launches into clear scopes, risk checks, and client-ready plans.