Instinct Is Worth $10 Billion Six Weeks After It Was Worth $2.5 Billion. The Product Is Still Behind a Waitlist

Instinct closed a $1 billion Series C on Monday from Sequoia Capital, Benchmark and Coatue at a $10 billion valuation. That is the headline number, and it is the less interesting one. The interesting number is six weeks — the gap between this round and the last one, which valued the same company at $2.5 billion.

Reuters puts the August round at roughly $250 million, a figure founder Noah Shinn gave the Wall Street Journal at the time. Nothing shipped in between that the public can evaluate. Instinct is still in early access, and the company's own announcement ends by pointing readers at a waitlist.

What it is

Instinct — legally Spear Street Technology Inc., founded in 2025 by Shinn, previously of the customer-service AI company Sierra — sells a personal agent that runs errands. You text it or call it. It plans trips, orders groceries, cancels subscriptions, books restaurants and doctor's appointments. The design choice that separates it from a chatbot with tools bolted on is that it operates its own phone number and its own computer rather than acting inside yours.

Three things in the announcement are worth naming. Instinct Concierge is a white-glove tier for high-touch requests the agent does not attempt alone. The Trusted Person Network lets one user's agent coordinate directly with another user's agent, so two people's schedules get negotiated without either person in the loop. And an iMessage integration reads location to make suggestions. The company says execution is sandboxed and that it runs hallucination detection over the agent's output.

That last sentence is the company describing its own safeguards. No third party has tested them, because there is not yet a product at scale to test.

We're building Instinct to be the best personal agent that can handle the deeply personal nuances of everyday life. This funding helps us bring Instinct to more people and continue building the future of personal AI.

What $10 billion is buying

Not revenue: none disclosed. Not users: none disclosed. Neither the press release nor any of the reporting carries a single usage number.

The bet is legible anyway. If a personal agent becomes the layer people book, buy and schedule through, whoever owns that layer sits between consumers and a large share of consumer spending. Sequoia and Coatue are not underwriting Instinct's current reliability. They are underwriting the possibility that the habit forms here first.

Competition is why the price moved this fast. SiliconANGLE counts roughly a dozen platforms chasing the same consumer — Wajo, Ollie and Poke among the startups, plus OpenAI's ChatGPT agents, Meta's Muse and Apple's Siri. Every one of those already has more distribution than a waitlist. In a race decided by who people form the habit with, being early is worth more than being correct, and capital is the only way to be early at scale.

The gap this round does not close

September has not been a good month for the claim that agents are ready to hold real credentials. In the same week this round closed, researchers documented the JADEPUFFER-linked campaign in which attackers used compromised Azure service principals to delete cloud resources. That is an attack story rather than a consumer-agent story, but it sits on the same fault line: an agent with live credentials acting on live systems is only as contained as the narrowest permission somebody remembered to set.

Instinct's product is the consumer version of that exposure. An agent holding your calendar, your email, your location, your payment methods and its own phone line is by construction the most privileged software in an ordinary person's life. Sandboxing and hallucination detection are the right two things to claim. Whether they hold under load is a question early access has so far prevented anyone outside the company from answering.

There is nothing improper about that. Early-access products are supposed to be unproven, and investors are entitled to price a category rather than a cash flow. But it is worth being precise about what happened this week: a 4x markup was applied to a company on the strength of what its agent is expected to become, six weeks after the last markup, with no public evidence in between.

What to watch

The number that settles this is not the next valuation. It is the first month Instinct publishes how many people use the agent and how often it finishes a task without a human stepping in. A $10 billion price on a waitlist is a statement about the category. A completion rate would be a statement about the company.

Sources: Instinct press release (Business Wire) · Bloomberg · Reuters · TechCrunch · SiliconANGLE · The Hacker News on JADEPUFFER