Instinct Raises $1B at $10B Valuation With Only 14 Employees

A $1 billion round for an invite-only assistant

Instinct, a San Francisco startup that has been in business for about a year, has raised $1 billion in a Series C round at a $10 billion valuation. Sequoia Capital, Benchmark Capital and Coatue co-led the round.36 The company has 14 employees, and its product is still available only by invitation.3 The valuation is four times what investors paid a month ago. Several outlets report that Instinct raised $250 million at a $2.5 billion valuation the previous month.568

Founder and CEO Noah Shinn says the money will let more people use the product and keep development going.2 One detail in the reporting changes how the deal should be read: most of the new money is expected to go toward buying compute.35 One widely shared reaction online put it bluntly. The round is less a bet on 14 people than a prepaid GPU bill that comes with a team.7

So the case for the deal rests less on the 14-person headline and more on what $1 billion buys, plus whether Instinct's unusual growth numbers hold up once access opens.

What Instinct actually sells

Instinct is a personal AI agent that carries out errands and tasks instead of just answering questions.2 It has no standalone app. People reach it by text, phone call or email, and the agent uses its own phone and computer to do the work.36 Shinn has described it as an assistant that can act on its own initiative. For urgent matters, it can call the user, such as when a document needs signing minutes before a deadline.3

Its examples include planning trips, ordering groceries and cancelling subscriptions.6 Two newer features expand that. Instinct Concierge handles jobs that need phone calls or service bookings. The Trusted Person Network lets two users' agents talk to each other to set up meetings and share files.26 Coverage of the company's own messaging says it stresses safeguards such as isolated sandboxes, short-lived credentials, and a check meant to catch subtle hallucinations before the agent acts.9

Shinn is 23 and was previously a research scientist at Sierra AI.3 One report says he dropped out of university to write research papers on AI agents before building the product himself.8 The assistant launched in August.8

The numbers behind the price

The sharpest split in the coverage is over what Instinct has disclosed about its business. Citybiz says the company has not released revenue or other financial measures that would let investors weigh the valuation against its current business.6 Under30CEO also notes that monetization details have not been shared.9

Other reports, drawing on a podcast interview with Shinn, include several operating metrics:

• Transaction volume: annualized transaction volume has passed $1 billion, and about half of it is travel.35
• Growth: the user base grows about 10% to 11% a day, with essentially no marketing spending.35
• Trust: about 40% of users give the agent a personal credit card within three weeks of signing up, and retention among users who share sensitive information is about 80%.35

These figures do not contradict the "no revenue disclosed" line, but they measure different things. Transaction volume is money flowing through the agent, not money Instinct keeps. The company says it does not want advertising and plans to charge transaction fees, which Shinn compares to Apple Pay: free for users, with merchants paying for distribution.3 He has pointed to a wide range of take rates, from about 2% to 3% at Shopify up to 30% for some boutique hotel bookings and Apple's in-app purchases. He says Instinct wants to be at the high end, but the company has not chosen a number.3 Until it does, the $1 billion volume figure is evidence of demand, not of profit.

The growth figure also needs context. Instinct first gave the product to about 200 people, mostly friends and family, each with five invites. Growth has compounded through those invites since then.3 Invites have reportedly sold on eBay for about $300.3 Flux points out that Clubhouse invites also sold for hundreds of dollars in 2021, and the app later fell from a valuation of roughly $4 billion to near irrelevance once anyone could join.8 Scarcity gives a small team time to fix bugs, and it also creates hype. The real test is retention after the waitlist ends.8

Why compute is the real story

Shinn has said that about 40% of his time goes to thinking about compute.3 Another report says that demand doubles every week.5 His concern is lead time. Compute orders take months to fill, so at 10% daily growth a company would have to decide now whether to buy capacity for a user base that could reach 100 million in three to four months.3 Instinct's workload is also unusual. The agent can wake early to prepare for a user's day, which creates background demand that coding assistants do not have.3 The company says separating background jobs from real-time conversations makes the same hardware three to eight times more efficient.3

This is why I read the round as a capital-intensity deal rather than a celebration of a small team. Commentators have calculated about $714 million of valuation per employee, which makes for a striking headline.7 But a consumer agent growing this fast mainly needs inference capacity, and that costs a lot no matter how many people are on staff. The investors appear to be paying for the right to buy that capacity before competitors do.

Not every claim in the coverage holds up. One widely shared post called Instinct "the most valuable AI startup in the world."7 That is not credible when the same post compares it with OpenAI and Anthropic. The comparison only works per employee, not by total value.7 At least one outlet covered the deal without naming the company and placed it inside 2024 funding trends.4 It is better to rely on the specific, sourced reporting.

The giants closing in

In terms of competition, the timing is risky. Meta's AI agent Muse launched on September 8 and passed 3 million downloads in the U.S. and Canada in just over two weeks, according to Sensor Tower data.3 A JPMorgan analyst called Meta "the most formidable competitor" among consumer AI agents, pointing to the commercial data behind its ad business and a planned hardware device.3 Alphabet has relaunched an upgraded experimental agent, and Manus expanded its agent lineup in the same week as Instinct's round.3 Meta's market value is about 191 times Instinct's new valuation.3

Other observers have asked why OpenAI, which made ChatGPT a consumer hit, has not shipped a product like Instinct. One guess is that competing with Anthropic for enterprise customers pulled its attention away.10 Another reply argued that OpenAI needs revenue soon to justify its spending and possible IPO, while Instinct is following a different strategy.10 Either way, Instinct now has an opening that the large companies are clearly trying to close.

Trust is both Instinct's main advantage and its main weakness. In August, the company faced a public backlash over terms of service that reportedly gave it "perpetual and irrevocable" rights to use user data. It revised the terms quickly.3 For an agent that holds credit cards and inbox access, another mistake like that could hurt it more than any competitor could.

What the round says about venture funding

Yahoo Finance described the deal as the debate over AI valuations "in its most extreme form." Investors are still assigning huge value to companies that could become the main way people use agentic AI, even as scrutiny of the sector grows.2 Under30CEO said investors are paying for position rather than revenue. The company people trust with everyday tasks could sit between them and many other services.9

I think that reasoning makes sense as long as the trust metrics hold up at scale. Four things will show whether the bet works:

• user growth after the invite system ends
• how reliably the agent completes tasks
• the take rate Instinct actually gets from merchants
• how much each active user costs to serve as compute spending rises29

If personal agents become the main way consumers deal with businesses, a $10 billion price could look cheap. If adoption slows, the gap between the valuation and the company's commercial size will be hard to justify.2

For now, the round shows that top-tier venture firms will pay frontier-lab-style prices for a consumer AI company that is a year old and has no app. Those valuations are increasingly being set by how much compute a company can secure, not by how many people it employs.