In a conversation with Patrick O'Shaughnessy, Instinct founder Noah Shinn said roughly 50% of the transaction volume through his company's AI agent comes from travel. That figure has a specific denominator: transaction volume on one invite-only platform. It does not mean half of all consumer AI searches are about travel.
Founder interview, starting at the travel discussion. Watch on YouTube if the player is unavailable.
What we see in real AI conversations
A read of Aiso's conversation dataset shows the questions that come before a travel purchase. These are paraphrased examples of prompts observed in our Supabase conversations table:
- Build a kid-friendly two-day itinerary for Chicago.
- Plan a short Bali visit around a noon arrival and an evening departure.
- Find Kansas City hotels convenient to several specific places, including a stadium.
- Map out a driving itinerary for Iceland in October.
- Suggest a Porto itinerary with food, architecture, nature, and history.
These are illustrative examples, not a random sample or a travel-share estimate. We selected English-language travel prompts on September 30, 2026 and paraphrased them to protect the original wording. The current topic labels do not isolate all travel searches reliably enough to calculate a defensible share of the full dataset.
The traveler is asking an assistant to combine constraints that a conventional search box often splits across several pages. A hotel query may include location, budget, nearby venues, and who is traveling. An itinerary request may include arrival time, children, driving distance, and interests. The answer can narrow the shortlist long before the person reaches a booking site.
What travel brands should do
Start with the decisions travelers actually ask an assistant to make. For hotels, make location, accessibility, transport, family suitability, and nearby points of interest easy to verify. For destinations and tour operators, explain who an experience suits, how long it takes, the logistics, and the tradeoffs. Keep prices and availability current in systems that support booking, and distinguish those live facts from evergreen guidance.
Then inspect the prompts where your brand could be relevant. What constraints appear? Which alternatives does the assistant name? What sources support its answer? Aiso helps brands find these real questions and see where they appear in AI answers, so content improvements can address a specific decision rather than a generic travel keyword.
Instinct's figure points to substantial travel commerce on one AI agent. Our examples show the earlier research behavior: travelers are already using AI to shape the shortlist. For a travel brand, the opportunity begins before checkout, when the assistant is helping a person decide what fits.
Source and scope
- Noah Shinn interview with Patrick O'Shaughnessy, travel discussion around 18:35. The approximately 50% figure is Shinn's statement about Instinct transaction volume, not an independently audited industry estimate.
- Aiso Supabase public.conversations, read September 30, 2026. Examples paraphrase records 646642, 755379, 628269, 591395, and 656792. These records support the examples, not a claim about the fraction of all assistant searches involving travel.
For broader travel context, see our travel conversation analysis.