Instinct Shipped a Text-Based AI Agent in August. Muse and Dots Came Next.

Instinct shipped an invite-only AI agent (autonomous task-completion software) in August and, within weeks, had become what The Verge called "the buzziest thing in AI." The startup's playbook — "invite-only, no marketing, and barely so much as a website" — violated nearly every convention of consumer AI launches. The Verge's reception turned that constraint into the story.

The product runs through SMS — text messages over standard cellular networks — bypassing the desktop dashboards and mobile apps that dominate consumer AI. Users text a request; the agent executes; the user receives a confirmation. Wired tested the system and ran the headline "I Finally Found an AI Agent Worth the Risk." On X, early users @pitdesi and @srcasm posted demos of Instinct booking a California DMV appointment and dispatching a follow-up email, respectively.

The Verge's column opens with the line "Before there were cute little guys, there was Instinct." The article links that phrase to a parallel Verge column filed under the slug "meta-muse-ai-cute." The framing positions Instinct as the prior art: text-first, task-first, distribution-light.

What did Instinct actually launch?

An invite-only agent with three observable properties, all drawn from the source:

• Distribution. Invite-only, no marketing, no website beyond a single signup page.
• Interface. Text messages in, text messages out. No chat window, no model picker, no app install. The Verge describes it as a "straightforward, text message-based interface."
• Operational scope. Booking California DMV appointments, sending follow-up emails, handling multi-party email threads.

The Verge's column emphasizes what Instinct did not do. The startup did not advertise, did not price, did not push for mass adoption. Its only growth channel ran through referrals among existing users — a constraint that produced press coverage but, on the source's evidence, did not produce installed base at category-incumbent scale.

What changes when Muse and Dots enter?

The competition arrived inside the same column. The Verge writes: "Then Muse arrived, followed not long after by Dots. The same products, more or less, from two far more powerful companies."

That sentence carries the geometry. Three products enter the same task-completion category, with two operated by firms the source describes as "far more powerful" than Instinct's parent — a phrasing that places both entrants in the Big Tech tier. The Verge does not name Dots' operator in the column body. A slug embedded in the source — "meta-muse-ai-cute" — points toward Meta Platforms as Muse's operator.

The size differential, not the feature differential, is the source's emphasis. The Verge does not compare features, pricing, or reliability across the three products. It compares reach: a closed SMS rollout against two vendors with mass-market surface area.

Why does the collision matter?

Three framings from the source map the stakes:

The interface thesis. The Verge led with what made Instinct distinct: its plainness. Text-message in, text-message out. If Muse and Dots replicate the surface but ship at scale, Instinct's minimalist design becomes a feature set rather than a market position — a competitive liability rather than a differentiator.

The trust dimension. The Verge calls the agent's launch "buzzy." "Buzzy" is a runway, not a moat; it requires sustaining. Two larger entrants reset the trust conversation the moment they announce. Instinct's invite-only rollout, by the same logic, cannot broadcast reliability at scale — it can only demonstrate it inside a small cohort.

The unfinished clause. The Verge's column ends mid-sentence: "With Big Tech players suddenly in the mix, it was looking dubious that the startup's buzzy launch could keep …" The columnist declined to resolve the question. That editorial restraint signals the open window, not the answer.

What happens next?

The source does not project. Three trajectories are consistent with what the column does state:

• Acquisition. A larger operator absorbs the team for its interface-design record and installed base. The asset, on the source's evidence, is credibility in a category that has otherwise struggled with reliability.
• Hold. Instinct absorbs the competitive pressure and watches Muse and Dots surface their own failure modes — the dropped bookings, the hallucinated confirmations (an AI returning a confident but incorrect answer), the privacy disclosures — at scale. The startup catches defectors.
• Narrow scope. Instinct remains a closed rollout, contracting rather than expanding, possibly pivoting toward enterprise or white-label deployments (unbranded software resold under another vendor's name). Wired's headline framing — "worth the risk" — implies a category that has historically under-delivered; contracting the surface reduces that exposure.

The Verge's column ends without resolution. The startup's next public move will land in follow-up reporting on this beat.

Marcus Bennett covers AI infrastructure and the agent stack for AI In Context.