Instinct, August's Most Talked-About AI Agent, Now Faces Muse and Dots

In just a couple of months, Instinct went from a closed launch to a product that has to defend itself against two competitors at once. In August, the startup released its AI agent by invitation only, with no marketing and barely a website. That was enough to make it one of the most talked-about things in the industry. Then Muse appeared, and Dots followed shortly after. According to The Verge, these are more or less the same products, but from far more powerful companies.

How Instinct Took Off Without Advertising

The bet was on simplicity. The agent lives in text messages: no new app, no dashboards, no workflow setup. If you need something done, you write it, and the agent gets to work. As The Verge put it, it is always just one message away.

That interface proved more persuasive than any presentation. Users on social media shared concrete cases: the agent booked appointments at the DMV (the US motor vehicle and registration agency) and sent follow-up emails after meetings. These are boring tasks, but they are exactly what people want to offload to AI. Wired singled out the product in a piece about finally finding an agent worth taking a risk on. In a category where trust in autonomous actions remains the main barrier, that kind of praise is worth a lot.

The launch format also worked in its favor. Closed access and the absence of advertising created a sense of discovery: the people talking about the product were those who already used it.

What Changed with Muse and Dots

The problem is that the idea turned out to be easy to replicate. Muse, judging by an internal link in The Verge's article, is tied to Meta. Dots came out soon after it. The available portion of the article gives no details on their capabilities, but the point of the comparison is clear: similar functionality, vastly different resources.

Big companies have what a startup, by definition, lacks: a ready-made audience, money for compute, teams for integrations, and the ability to ship a product as a default right where people already are. Against that backdrop, the invite-only model that helped Instinct in August becomes a constraint. While the startup hands out access manually, its competitors can open the doors to everyone.

That is why the question in The Verge's headline is not rhetorical. In AI agents, a first-mover advantage is measured in months, not years.

Key Facts

• Instinct launch: August 2026, invite-only access
• Marketing: practically none, minimal website
• Interface: text messages, no separate app
• Example tasks: booking DMV appointments, follow-up emails
• Competitors: Muse and Dots from larger companies
• Open question: whether the startup can retain users once alternatives are available

Where the Startup Still Has a Chance

The first thing that may help is trust. An agent that acts on a person's behalf gets access to email, calendar, and personal data. Here, a small team with a clear product sometimes looks more trustworthy than a corporation whose advertising business runs on that same data. But this is a hypothesis, not a fact: the article contains no public information on how Instinct stores data or what access rights it has.

The second is iteration speed and focus. Big companies often ship an agent as one feature of an ecosystem, while a startup can polish a single use case until it is stable. For tasks like DMV bookings, where a mistake means lost time, reliability matters more than breadth of capabilities.

The third is habit. If someone has been texting with an agent for a month and a half and it knows their preferences, switching to another service takes effort. That said, in a messaging app switching costs almost nothing, which works against Instinct.

What Users and Teams Should Check

If you are choosing an agent for work or personal tasks, do not go by the buzz. Run three or four everyday scenarios on each service: booking an appointment, a follow-up email, a reminder with context. Compare how many times the agent asks clarifying questions and how many times it makes mistakes.

Also look at access permissions: what exactly the service connects to, and whether you can restrict its actions, for example by forbidding it to send emails without confirmation. For businesses, longevity matters too: will the vendor still exist in a year? With a startup in this position, it makes sense not to build critical processes on it until it becomes clear how it will respond to the arrival of Muse and Dots.