OpenAI's Dots Agent Targets Enterprise Tasks, Not Consumer Chat

What happened

OpenAI has introduced Dots, a new agent platform designed to carry out tasks on behalf of users inside ChatGPT, spanning both workplace functions and everyday errands such as ordering food. According to hands-on reporting from The Verge, the product presents itself primarily as enterprise software, with personal conveniences like food ordering positioned as a secondary capability rather than the core pitch.

The framing marks a contrast with Meta's recently discussed "Muse" agent, which reporters describe as leaning into a more approachable, consumer-facing personality. Dots, by comparison, reads closer to conventional workplace software that happens to carry out personal requests on the side, rather than a chatty assistant built for everyday life.

Why it matters

The positioning signals where OpenAI sees the near-term commercial opportunity for agentic AI: inside organisations, handling multi-step work tasks, rather than as a lifestyle companion competing for everyday consumer attention. For enterprise and digital transformation leaders, this suggests agent platforms are moving from novelty demonstrations toward tools expected to plug into existing workflows and systems of record.

It also reflects a broader split emerging in the agent market between consumer-oriented personalities designed to build emotional engagement, and enterprise-oriented agents designed to build trust through reliability and task completion. How a platform presents itself — playful companion versus efficient tool — shapes user expectations, adoption patterns and the kind of behavioural trust organisations need to build with employees and customers alike.

The Renascence take

The real story here isn't the burrito order — it's the design choice behind it. OpenAI could have leaned into a friendly, consumer-style persona the way Meta has, but chose a more utilitarian tone, and that choice sends a clear signal about who the product is really for.

Most commentary will fixate on the novelty of an AI agent fetching dinner, but the more telling detail is the personality OpenAI chose not to build. Enterprise users don't primarily need charm from their software; they need predictability, and predictability is itself a trust-building design decision, not an absence of design. Organisations piloting agentic AI should resist the urge to make these tools "likeable" by default and instead ask what tone actually earns confidence from the people who'll depend on them daily. The companies that get this right will treat agent personality as a deliberate service-design variable, not an afterthought bolted on for demos.