AI startup Instinct has raised $1 billion in a Series C round, lifting its valuation to $10 billion. Behind the eye-catching numbers is an AI direction investors are betting big on: instead of just answering questions, an AI agent can take a request by text or call, then use its own phone and computer to complete a task end to end.
From $2.5B to $10B in about a month
Instinct announced the new round on September 28. The $1 billion Series C included three big names: Sequoia Capital, Benchmark Capital and Coatue. After the round, Instinct is valued at $10 billion. The striking part is the pace of the markup: per reports, the prior Series B raised $250 million at roughly a $2.5 billion valuation.
| When | Round | Amount | Valuation |
|---|---|---|---|
| Aug 2026 | Series B | $250M | ~$2.5B |
| Sep 2026 | Series C | $1B | $10B |
So Instinct's valuation rose about 4x in roughly a month — very fast even in an AI market awash with capital.
What does Instinct actually do?
Instinct describes itself as a personal AI agent for everyday life. The difference from a traditional chatbot is the word 'do'. A chatbot might answer 'here are some hotels for your trip'; Instinct aims to take a request — 'plan this trip for me' — and handle the many steps that follow. Per the company, users just text or call Instinct; the agent has its own phone and computer to carry out the work.
Examples the company gives include: planning a road trip, ordering food and groceries, buying tickets, managing or canceling subscriptions, booking services, and handling other administrative tasks.
From chatbot to AI agent
This is why Instinct is more notable than a new chatbot startup. Simply put: with a chatbot, you ask → the AI answers → you do it. An AI agent aims for: you assign a task → the AI plans → the AI uses tools → the AI does multiple steps → the task is done.
For example, to book a restaurant that doesn't take online reservations, a chatbot might give you the phone number; a more ambitious AI agent would call the restaurant itself and handle the booking. That's exactly the kind of situation Instinct is building for.
Instinct Concierge can call on your behalf
One new product the company introduced is Instinct Concierge, aimed at tasks needing more interaction: making calls, booking premium services, reserving at restaurants that don't take online bookings, joining a dentist's waitlist, or handling a cable-TV bill. It's a notable step: AI no longer lives only in a chat window but starts entering workflows that used to need a person — open a website, find info, call, wait, discuss, confirm. If this model works reliably at scale, the idea of a 'personal assistant' could become far more real than today's chatbot generation.
Your AI can talk to your friend's AI
Instinct also introduced a Trusted Person Network. Per the company, it lets one person's Instinct communicate directly with another person's Instinct to coordinate plans; the agents can also share files like PDFs, images and spreadsheets. A future example: you say 'find a free day this weekend to hang out with An' — instead of you checking your calendar, texting An, An checking theirs, texting back, rescheduling and re-checking, the two agents can work out the logistics and propose a suitable time. It's an interesting direction for agentic AI: AI not only using tools, but starting to coordinate with other AI.
Location Sharing brings AI closer to real-world context
Instinct added the ability to receive a user's shared location via iMessage. Per the company, the agent can recognize when a user arrives somewhere and offer suggestions based on preferences or answer questions relevant to that context. This reflects a broader trend: first-gen AI assistants mostly knew what you just typed into chat, while the next generation of personal agents wants more context — calendar, email, messages, location, preferences, the services you use.
The more an AI can do, the more sensitive its access
For an agent to buy things, cancel subscriptions, make calls, use accounts and read files, it needs more permissions than a regular chatbot. Instinct says the product uses measures like isolated sandboxes, short-lived local credentials and identity-signed tool execution; it also says it built a system to detect subtle AI errors before the agent responds or acts. This will be one of the biggest problems for personal AI agents: an AI is more useful the more access it's given — but the more access, the bigger the consequences of a wrong decision.
Instinct is still in Early Access
One detail to keep amid the huge valuation numbers: Instinct is still in Early Access, not a consumer product already rolled out to hundreds of millions. The Information (before the round) reported that Instinct's user count had passed 100,000, while fast-rising demand had at times strained the service's compute — this is from press reporting, not a figure the company announced in its Series C release. That makes the $10 billion valuation all the more notable: investors are betting not just on current scale but on personal AI agents becoming a large product category in the future.
When was Instinct founded?
Instinct was founded in 2025 by Noah Shinn, who previously worked at AI customer-service startup Sierra; the company is based in San Francisco. Instinct began testing the assistant with a small group of users in early 2026 and later drew attention as users shared what the agent could do. In a relatively short time, the startup went from private testing to a $10 billion valuation.
Why are investors so interested in AI agents?
The first wave of generative AI focused heavily on writing, Q&A, image generation, coding and search. The agentic-AI wave shifts the question to: what can AI complete on its own? Instead of creating a travel itinerary, an agent can book the trip; instead of explaining how to cancel a subscription, the agent can cancel it; instead of finding a dentist's phone number, the agent can call. That's why big tech firms and startups are pouring resources into agents; per Reuters, Instinct's round comes as investor interest in agentic AI keeps rising sharply.
But a $10B valuation doesn't mean Instinct has proven the model
This needs to be kept separate. A valuation reflects the price investors accept in a round; it doesn't mean $10B in revenue, $10B in profit, or a product that has succeeded at mass scale. Instinct is still in Early Access. AI agents generally still have to solve many issues: accuracy, privacy, security, action permissions, compute cost, and resilience when a step in the workflow fails. So the most notable thing about the round isn't 'Instinct has won', but that investors are willing to bet heavily on the idea that AI can shift from 'answering' to 'acting'.
30-second recap
- How much did it raise? — $1 billion in a Series C round.
- Valuation? — $10 billion.
- Who invested? — Sequoia Capital, Benchmark Capital and Coatue.
- Prior valuation? — About $2.5 billion in a Series B a month earlier.
- What does Instinct do? — A personal AI agent that takes requests by text/voice and uses a phone and computer to do everyday tasks.
- Examples? — Planning trips, buying groceries, canceling subscriptions, booking services, handling call-based tasks.
- Widely available yet? — No. Instinct is still in Early Access.
