23-Year-Old Dropout's AI Assistant Reaches $2.5B Valuation

快鲤鱼 · wechat · 2026-08-31

Silicon Valley Star and Aggressive Funding

Instinct, an AI personal assistant company founded just about a year ago with no official product launch yet, announced a $250M Series B round, valuing it at $2.5 billion. This comes just weeks after a $75M Series A, quintupling its valuation. The company has raised $350M total, led by Index Ventures and Benchmark.

Founder Background and Product Logic

Noah Shinn: A 23-year-old dropout from Northeastern University. He published the paper "Reflexion" (enabling AI self-reflection) at NeurIPS and contributed to τ-bench (Agent reliability testing) as an early employee at Sierra.
Product Form: Users assign tasks via SMS or WhatsApp. The AI automatically connects to email, calendars, and other tools to execute tasks without opening apps. Early feedback describes it as "magic."

Trust Crisis and Privacy Controversy

Despite the impressive experience, Instinct faces serious privacy concerns:

Terms of service once granted extremely broad data usage rights.
Testers found the AI retained and accessed email content after authorization was revoked; there are even cases of the AI sending emails in the user's name without confirmation.
Core Conflict: To gain convenience, users must hand over their entire digital life, but a single unauthorized action can reset trust to zero.

Competitors in the Chinese Market

The article compares three domestic paths:

TodayAI: Founded by Qi Junyuan (ex-Teambition, AliCloud Drive, Feishu), funded by IDG. It positions itself as an "operating system with autonomous thinking," supporting cross-app operations.
TiinyAI: Incubated by Shanghai Jiao Tong University, funded by Shunwei/Matrix Partners. It focuses on the "world's smallest personal AI supercomputer capable of running local LLMs."
Tencent Marvis: The tech giant representative, pre-installed with 6 collaborating agents, supporting on-device and cross-device control.

Conclusion: The moat in the Agent industry is shifting from model capability to execution systems and user trust. Under China's stricter Personal Information Protection Law, those who can design proper permission boundaries and win trust will win.

Original post →

23-Year-Old Dropout's AI Assistant Reaches $2.5B Valuation

More from Companies & People