This insight summarises From $0 to $10 Billion in Months: Silicon Valley's AI Assistant Gold Rush Is On (and What It Means for Travel), by Ivana Johnston, co-founder and CEO of Puzzle Partner, a B2B marketing, PR and go-to-market firm serving SaaS, hospitality and travel companies, published on Hospitality Net on 12 November 2026.
The headline number. Instinct is an AI assistant people use by text or email rather than through an app. It launched its invitation-only assistant in August, and "in September, Instinct raised $1 billion at a $10 billion valuation." A month earlier, investors had valued it at $2.5 billion.
Why travel should pay attention. "According to its founder, more than 50% of Instinct's transactions involve travel." The company also reports approaching $1 billion in annualised transaction volume, a figure Johnston notes comes from the company. One user changed flights and secured a difficult restaurant reservation through a connected OpenTable account.
A new kind of race. AI companies are competing less on benchmarks and more to become "the assistant you trust to run parts of your life", which means access to calendars, inboxes, contacts and eventually money. Meta launched its Muse personal agent in September, with distribution through WhatsApp, Instagram, Facebook and Messenger. xAI's Grok Bot lets users build teams of specialised agents. Anthropic introduced Claude Duet, a proactive assistant in research preview, in October, and OpenAI's Dots offers persistent agents for ongoing projects.
Ten assistants, no ranking. Johnston reviews Instinct, Catch AI, Meta Muse, Claude Duet, Grok Bot, OpenAI Dots, Viktor, Lindy, Martin and Must AI Assistant, calling the list "my current interests and assessments, rather than a ranking based on identical hands-on tests." Instinct is Johnston's current favourite for simplicity, Muse for personal tasks and Catch for executive administration, with Claude Duet a possible leader for professional use.
Still imperfect. "Websites block automated interactions, instructions get misunderstood and occasional tasks require human intervention." The more an assistant learns about preferences and accounts, the harder it is to switch. Access to email, purchases and finances raises the privacy stakes. Meta, for example, runs Muse in a dedicated virtual environment and requires approval before certain purchases. Long-term winners, Johnston argues, will depend on reliability, privacy protection and whether users keep delegating more.
AIHA practical takeaway. If more than half of one assistant's transactions already involve travel, expect more guests to delegate hotel research and booking. Make rates, availability, policies and add-ons readable by machines on your own site. Check that your bot settings let legitimate assistants reach a booking path you control, and that your reservations team can serve a request placed by an assistant on a guest's behalf.
Related AIHA insight: Why the Personal AI Assistant Hype Matters More for Travel Than Any AI Travel Product.