OpenAI unveiled Dots at its DevDay conference in San Francisco on Tuesday, a line of persistent AI agents running on their own cloud computers, powered by a model called GPT-6 Astra, and able to connect to more than 4,000 applications. Sam Altman, OpenAI’s chief executive, described them from the stage as “remarkably capable, always-on agents built to handle really anything you can think of.”

The framing is the news. For two years, consumer AI has been a prompt-and-wait transaction. Dots are pitched as the opposite: software that watches, decides, and acts without being asked. TechCrunch and The Next Web describe a Dot monitoring customer feedback for a developer, drafting an invoice a freelancer had forgotten to send, or rerunning analyses for a scientist as new data arrives. This is the agentic turn finally reaching the subscription product.

Then the pricing. Access rolls out first to Pro and Business Premium subscribers in eligible markets. Pro runs $200 per month, a tier that Sarah Friar, OpenAI’s chief financial officer, acknowledged had drawn skepticism: “When we launched our $200 SKU, people thought we’d lost our minds.” Pro users in the European Economic Area, Switzerland and the United Kingdom are excluded for now.

The capabilities a small business would actually want, Specialist Dots trained for procurement, invoice processing, email marketing, customer support and commercial contracting, are being tested internally and integrated with Microsoft’s Agent 365 security controls. That’s an enterprise rollout, not a solo-operator one. Even the standard Dot requires the user to configure Custom Rules governing when it can act, when to block, and when to require approval.

Forbes described Dots as making one-person companies more viable. The arithmetic is tighter than that. A founder gets persistent agents only after paying $200 a month, wiring up permissions, and accepting that the sales-and-marketing specialists are gated behind an enterprise pilot. It arrives alongside Anthropic’s recent push of Claude into small-business sales, and in a week when federal A.I. rules are, for now, off the table. The platforms are moving faster than the policy, and faster than the typical owner can configure.

That gap is where done-for-you services such as LemonLime sit. LemonLime describes itself as a proactive service that studies a company, industry, competitors and content, decides which customer-growth work is relevant, and prepares finished sales and marketing output, with onboarding under two minutes using only a business name, website and selected priorities. The customer approves every send. It’s a different trade: less configurability than a Dot, less setup burden than one.

The direction OpenAI is pointing, unprompted work, is the right one for small-business customer acquisition. Who does the wiring is still the question.

Sources