It was the kind of week that turns a founder into a fixture. On Monday 28 September 2026, Instinct — the invite-only personal AI agent that runs entirely over text message and phone calls — announced a $1 billion Series C at a $10 billion valuation, backed by Sequoia Capital, Benchmark Capital and Coatue. Within a day, founder Noah Shinn was sitting down with investor Patrick O'Shaughnessy to explain what the company is actually building. TechCrunch reported the conversation on 29 September, and it offers the clearest window yet into how the buzziest startup of the agent era sees its future.
Meridian did not conduct this interview. This feature is an editorial reading of Shinn's published on-record remarks, with every quotation taken from the coverage cited in the sources below.
The numbers alone are startling. A 14-person startup, launched in invite-only early access with no mobile app and no disclosed revenue or user counts, quadrupled its valuation in about a month — from a $2.5 billion Series B to a $10 billion Series C. Investors are not buying metrics; they are buying a thesis: that an AI agent which acts across your apps, devices, email, messaging, screen, audio and location can become the primary interface for everyday life.
A billion dollars for the personal agent
In the company's funding announcement, Shinn framed the raise as fuel for a mission, not a milestone:
"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. It's an exciting, creative time, and we're just getting started."
— Noah Shinn, Instinct funding announcement, 28 September 2026 (via Business Wire)
The company's recent product updates sketch the ambition: Instinct Concierge, a white-glove service for high-touch cases like phone calls and hard-to-book services; a Trusted Person Network that lets one user's Instinct talk to another's to coordinate plans; and location sharing so the assistant recognises when you've arrived somewhere. The interface is deliberately minimal — you text or call, and Instinct uses its own phone and its own computer to complete the whole task, from planning a cross-country road trip to cancelling a forgotten subscription.
Travel is the killer app
The headline revelation of Shinn's conversation with O'Shaughnessy was where the agent's usage actually concentrates. Almost every AI demo shows an assistant booking travel — and Instinct's data says the cliché is real:
"People who are not on Instinct use travel agencies (or sites) to book hotels or flights. What's the benefit of that interface? They properly unify different hotel chains, airlines, and services, and present them in a nice way for users, who have the option to check out immediately."
— Noah Shinn, in an interview with investor Patrick O'Shaughnessy, as reported by TechCrunch (29 Sep 2026)
More than half of all transactions on the platform are travel-related, Shinn said, and the invite-only platform is approaching a billion dollars in annual transactions — growing, he claimed, at 10 per cent day by day. His pitch for why agents beat booking sites came down to the difference between browsing and delegating:
"I think there is a better experience with Instinct or other interfaces where users can say, 'Hey, I need to be in New York tonight; it's urgent.'"
— Noah Shinn, in an interview with investor Patrick O'Shaughnessy, as reported by TechCrunch (29 Sep 2026)
It is a neat encapsulation of the agent thesis: the booking site unifies supply and presents it nicely; the agent removes the interface altogether. You don't search — you state the emergency, and the machine does the rest.
Reinventing the restaurant reservation
Shinn's most provocative remarks concerned a far smaller transaction: the restaurant table. He argued that the first-come-first-served reservation system ignores what both sides actually want — and that an agent checking restaurant sites every few seconds could reinvent the whole ritual, even giving users preferential treatment:
"If you take restaurant reservations from scratch, from the user's end, they want the reservation for a significant life event that they might have. And on the restaurant end, they just want interesting people. They want special occasions. They want to cater to those rather than, maybe, the local [person] who keeps, you know, taking up a table once a night, and there's no sort of special event there."
— Noah Shinn, in an interview with investor Patrick O'Shaughnessy, as reported by TechCrunch (29 Sep 2026)
The comments drew immediate pushback. As TechCrunch noted, observers pointed out that agents can simply lie about anniversaries and special occasions to secure tables, and that many restaurants genuinely prefer their local regulars — the customers who bring steady business over the trophy diners who visit once. The exchange exposed the central tension of the agent economy: a system optimised for the user's special occasion may be optimised against everyone else's. Instinct's answer, so far, is that the interface itself is due for reinvention — but the social contract of the dining room may prove harder to reprogram than the software.
Who is Noah Shinn
Shinn's path to a $10 billion valuation runs through research labs, not business schools. He was previously a research scientist at Sierra, and conducted machine-learning and programming-language research at Northeastern and MIT. As a Northeastern student, he co-developed Reflexion, a framework that has a language model check its own output and use feedback to improve subsequent attempts — work reported to have reached 91 per cent accuracy on the HumanEval coding benchmark, 11 percentage points above the result then attributed to GPT-4.
That background is a useful clue to Instinct's engineering philosophy. Reflexion was about getting an AI system to recognise its own mistakes and recover from them — which is, in miniature, the entire problem of a personal agent: doing a task is one thing; noticing you did it wrong and fixing it is another. An assistant that books your flights, calls your dentist and coordinates with your friends' assistants cannot afford to fail silently. Shinn's research career was, in effect, a rehearsal for the trust problem his company now has to solve at scale.
The agent wars
Instinct is not alone in the race. TechCrunch notes the 14-member startup is competing with Meta's consumer AI agent Muse and with Wajo, the Khosla-backed entrant in the personal-agent space. What distinguishes Instinct is its radical interface minimalism: no app, no new surface to learn. The agent has its own phone number and its own computer, and it acts through them the way a human assistant would.
Whether that model survives contact with the messy economics of travel commissions, restaurant relationships and agent-to-agent coordination is the open question of 2026's AI funding frenzy. For now, though, Shinn has something few founders ever get: a week in which the money, the metrics and the narrative all arrived at once. The hard part — turning 10 per cent daily growth into a durable business — starts now.
Sources
