OpenAI's new always-on agents could overlap with the automation tools small teams already pay for. Here is what has been reported, what hasn't, and how to think about it.
According to coverage from VentureBeat, TechCrunch and others, Dots are persistent agents that work continuously in the background rather than waiting for a prompt. Each one reportedly gets its own cloud computer and browser, can be given a custom identity, credentials and tools, and can collaborate with other Dots. They launched alongside ChatGPT Space, a shared workspace that replaces the Library feature, where employees, ChatGPT, Codex and Dots work from shared files and context.
Reported integrations include Slack, Microsoft Teams and Google Drive, plus 4,000+ apps through OpenAI's plugin ecosystem. Microsoft Agent 365 security integration is described as still in development.
Per the reports, Dots are available in ChatGPT on web and mobile for Pro and Business Premium subscribers in eligible markets, with the first Dot included at no additional charge. Enterprise, Edu and Healthcare plans get beta access when an administrator enables it. What additional Dots, specialist Dots or extra capacity will cost has not been disclosed. We are not quoting dollar figures here because none have been confirmed, and the details could change. Check OpenAI's own pricing pages before budgeting.
OpenAI says background activity is limited to read-only operations, that custom rules can allow, require approval for, or block actions, that sensitive actions such as password changes need human approval, and that Business and Enterprise data is not used for training by default. OpenAI also acknowledged that Dots can make mistakes and that consequential work should be reviewed. A live demo agent reportedly paused for roughly ten seconds before answering a status question, a small moment but a reminder that "always-on" is not the same as "always reliable." TechCrunch also noted overlap with existing Codex capabilities, suggesting part of this is repackaging.
These are different categories, and the tradeoffs matter for a small business.
| Question | Dots (agent) | Zapier / Make / n8n (workflows) |
|---|---|---|
| How work is defined | Goal in plain language; the agent decides steps | Explicit triggers, steps and branches you design |
| Predictability | Lower; review of consequential output is advised | Higher; same input, same path |
| Best fit | Monitoring, research, drafting, open-ended tasks | Repeatable processes: lead routing, invoices, reporting |
| Cost model | Subscription tier; extras undisclosed | Published per-task or per-operation plans (n8n also self-hostable) |
For a repeatable job like "new form submission goes into the CRM and triggers a follow-up," a deterministic workflow is still the safer, easier-to-audit choice. Agents are more interesting for fuzzy work: watching customer feedback, summarizing what changed overnight, preparing a first draft for you to approve. For where workflow tools stand, see our Make vs Zapier comparison and our automation tools roundup.
If you already pay for a plan that includes a Dot, try one low-stakes, read-only task first, such as a daily digest of support emails or competitor mentions, and keep the approval rules strict. Keep revenue-critical processes in your existing workflow tool until pricing, reliability and permissions are clearer. If you don't have an eligible plan, there's no reason to upgrade on launch-week news alone.
Note that StackPilot's comparisons are based on research of public information and vendor documentation, not hands-on testing. This article reflects press reports from launch week and will be updated as OpenAI publishes more detail.