Instinct Raises $1 Billion at $10 Billion Valuation in Series C
The personal AI-agent startup raised $250 million at a $2.5 billion valuation about one month earlier as Meta expands its Muse assistant in the U.S.
Instinct has raised $1 billion at a $10 billion valuation, giving Noah Shinn’s personal AI-agent startup fresh capital as Meta expands its own consumer agent, Muse, in the U.S.
The Series C was announced Sept. 28 at 8:03 a.m. ET (12:03 p.m. UTC). Sequoia Capital, Benchmark Capital and Coatue participated in the round.
The financing follows Instinct’s earlier $250 million Series B, which valued the company at $2.5 billion. Index Ventures and Benchmark co-led that round in August.
Instinct’s assistant can be reached by text or phone call. It uses a phone and computer to perform tasks including planning travel, ordering groceries and canceling subscriptions.
The company also announced Concierge, a feature designed to make phone calls and handle high-touch bookings. Other tools include a Trusted Person Network for coordinating and sharing files, along with location-sharing functions through iMessage.
Instinct says its systems use isolated sandboxes, short-lived local credentials, identity-signed tool execution and an “active detective system” designed to detect hallucinations before responses are generated or actions are executed.
“We’re building Instinct to be the best personal agent that can handle the deeply personal nuances of everyday life,” Shinn said. He said the new funding will help bring Instinct to more people and support development of its personal AI platform.
Meta introduced Muse on Sept. 8 as a personal AI agent that can work across connected applications, including sending email and booking travel. Meta says Muse runs through a dedicated “Muse Secure VM,” can request approval before sensitive actions and is rolling out in the U.S. on iOS, Android and the web.
Instinct’s website describes a system connected to email, messaging, screen, audio and location while operating through a user’s phone and computer. The service remains in early access.
Shinn previously co-developed Reflexion, a framework that enables language models to evaluate and improve their own responses. Northeastern University said the method reached 91% accuracy on the HumanEval coding benchmark, compared with 80% for GPT-4.
