UpTrajectory Review

OpenAI used its 2026 DevDay stage to introduce Dots, a new class of always-on AI agent built on the company's GPT-6 Astra model. Each Dot is a persistent, autonomous assistant with its own cloud computer, accessible through ChatGPT, Slack, Teams, or voice calls. Unlike the chatbots most operators have experimented with, Dots are designed to run continuously in the background, learning from your feedback over time and proactively working toward goals you set. The launch comes barely a week after Meta unveiled Jolly, a cheerful mascot for its own Muse assistant, making the race to anthropomorphize AI tools suddenly feel crowded.

For a small-business operator, the shift from reactive chatbot to proactive agent is the real story here, not the cute packaging. A Dot that remembers project context across days and keeps tasks moving between meetings could function like a junior employee who never sleeps. The ability to message one inside Slack or Teams means these agents slot into workflows your team already uses, lowering the friction of adoption. If the technology works as advertised, solo founders and lean teams gain a kind of always-on operational capacity that was previously affordable only at larger headcounts.

The mascot strategy, however, deserves skepticism. Naming a tool Iggy or Felipe and giving it glasses or a bow-tie is not neutral design. It builds emotional familiarity and trust before the user has any real basis for judging the agent's reliability. That matters when the same tool is connected to your accounts, your communications, and your business data. OpenAI is betting that a friendly face will ease adoption, but the comfort an avatar provides can outpace actual competence, leading operators to delegate more than they should, sooner than they should.

There is also a competitive signal in the timing. Meta's Jolly and OpenAI's Dots arriving within the same week suggests both companies believe personified agents are the next interface battleground. For small businesses, that means the tools will keep improving, but it also means lock-in risk grows. A Dot that learns how you think and what good work looks like to you becomes harder to replace with each passing week. Choosing an agent platform is starting to resemble hiring: the onboarding investment is real, and switching costs climb over time.

Watch how OpenAI prices Dots and what guardrails ship with them at launch. Persistent agents with their own cloud computers will not be cheap to run, and the business model will shape whether this is a tool for enterprises or genuinely accessible to small operators. In the meantime, the practical move is to identify one narrow, low-risk workflow, such as meeting follow-ups or inbox triage, and pilot an agent there before expanding its reach. Trust the output, not the avatar.

Takeaway: Persistent AI agents could add real operational capacity for lean teams, but pilot them on narrow tasks and judge results, not the friendly avatar.