On September 22, at the 2026 TravelDaily Conference, Shaohua Li, former president of Alibaba’s travel platform Fliggy and now co-founder and CEO of Beijing Shilv Network Technology, shared his views on the next opportunities for AI applications and investment in travel during a keynote titled “The Next Opportunity in Travel AI Applications and Investment.”
If his view could be summed up in one sentence, it would be this: travel AI has entered the competition for execution.
Models are responsible for understanding user intent, while AI agents organize tasks. But actual fulfillment still depends on travel companies. AI is reshaping the travel service process, but it has yet to truly take over transactions and fulfillment.
The change becomes more tangible through two real-world tests.
During a trip to the U.S. this summer, Li used only ChatGPT-5.6, without relying on any third-party tools.
The first was a family trip from New York to the Washington, D.C. area, including the White House and Capitol Hill.
GPT generated an itinerary planned down to five-minute intervals and proactively ruled out his original plan to rent a car, recommending a train instead. In the end, he followed the itinerary almost exactly.
The only steps that still required human intervention were purchasing train tickets and attraction tickets. GPT could handle the planning, but users still had to click through to external links to make the purchases, as it could not place orders or make payments on their behalf.
The second was a 15-day inbound trip starting in New York and covering nine destinations, including Shenzhen, Guangzhou, Guilin, Chengdu and Shanghai. The trip coincided with the beta testing of an inbound travel product being developed internally by Shilv Technology.
As a travel industry professional with two decades of experience, Li said he could hardly find fault with the AI-generated itinerary. Instead, AI proactively flagged several easily overlooked risks, including tight train ticket availability during China’s National Day holiday, boat ticket arrangements in Yangshuo, and the connection between Chengdu and Shanghai.
This led him to a more direct conclusion: GPT has already rewritten the personal travel planning process.
In the past, travelers had to search for information, compare options and repeatedly revise their itineraries. Now, much of this work can be completed through continuous conversations with AI. Users only need to clearly express their intent, while research, planning and revisions can be delegated to AI.
The main gaps that remain are booking and payment.
In Li’s view, however, these are not insurmountable problems. The airline industry has already demonstrated that AI can eventually move deeper into the transaction process.
The biggest gap in travel AI may not be capability, but trust.
During his presentation, Li cited research comparing the Chinese and U.S. markets. China is the world’s second-largest travel market, while the U.S. market is approximately 35% to 40% larger. According to one survey he cited, 61% of U.S. consumers are already using agentic services, 83% of travel companies have adopted general-purpose AI capabilities, and 39% of consumers use AI for trip planning.
However, another survey found that only 8% of consumers actually complete bookings directly through AI.
There is still a significant gap between “using AI” and “entrusting AI with the transaction.”
