In a transformative step for the travel and artificial intelligence industries alike, Kayak and Expedia are revamping how we plan our trips—by turning social media inspiration into personalized itineraries using advanced AI agents. The new battleground is no longer a search box on a website; it’s Instagram grids, TikTok vlogs, and Reddit threads. With the rise of generative AI and multimodal large language models (LLMs), these platforms are leveraging user-generated content to craft hyper-personalized travel experiences, blending discovery and booking in a frictionless loop.
The Technology Converting Social Signals into Actionable Itineraries
Search engines dominated digital travel planning for over two decades, but in 2025, we see a fundamental shift in consumer behavior. Browsing social content is now the norm for destination discovery. According to a March 2025 Pew Research Center survey, 61% of millennials and 78% of Gen Z travelers discover travel locations through video content and short-form posts. Recognizing this behavioral pivot, Kayak and Expedia are implementing sophisticated AI models capable of parsing and understanding visuals and natural language from social platforms like Instagram, TikTok, and Pinterest.
According to Kayak CTO Matthias Keller, the company’s new AI travel companion can ingest a user’s social media content—such as an inspirational reel of Paris street cafés—and output a structured travel itinerary in minutes. This is enabled by integrating computer vision, natural language understanding, and mapping APIs into a multimodal AI stack. “We no longer need you to know where you want to go. We only need you to know what excites you,” said Keller in a recent interview with VentureBeat.
These AI agents blend machine learning with real-time pricing APIs, dynamic inventory databases, and geo-tagged metadata to recreate itineraries faithful to original inspiration, with real bookings and transport options attached. It’s not just about matching a video with a place—it’s about reinterpreting a mood board into a tangible travel experience using complex reasoning layers from LLMs.
Key Drivers of This Evolution in AI-Powered Travel
Shifts in Consumer Preference and Content Consumption
Recent studies by the World Economic Forum highlight how younger travelers are consciously avoiding traditional travel blogs or corporate travel bundle systems. Social curation has become the leading influence. Coupled with the fact that 1.2 billion people are forecast to be active on TikTok by mid-2025, Kayak and Expedia are investing in social listening algorithms to extrapolate interest signals from these platforms.
A significant trend revealed by Deloitte’s 2025 Future of Work Report reflects how digital natives demand tools that can make sense of large volumes of media without requiring manual research. The success of ChatGPT, Midjourney, and other generative tools has prepared users for AI-curated workflows everywhere, including vacation planning.
Advancements in Multimodal Foundation Models
The foundation enabling this leap is the continued progression of multimodal large language models. OpenAI’s GPT-5 and Google DeepMind’s Gemini 2.5, both launched in early 2025, now support seamless integration of image, video, and textual analysis. On platforms like OpenAI’s ChatGPT Pro, a user can upload a snapshot of a city skyline pulled from Pinterest, and the assistant will decipher location, theme, and activity type—including matching architecture and cultural context—producing an itinerary that includes local experiences nearby.
As of May 2025, NVIDIA reported in its developer blog that travel-based AI applications are a leading GPU usage case outside of enterprise workflow automation, signaling increasing monetary investment and demand in the consumer AI sector.
| Company | Model Used | New AI Travel Feature |
|---|---|---|
| Kayak | Proprietary Multimodal AI + OpenAI GPT | Transforms TikTok videos and user prompts into custom booking itineraries. |
| Expedia | Amazon Bedrock + Claude 3 (Anthropic) | Combines social media context scraping with adaptive itinerary building. |
Source technologies such as Amazon Bedrock, Midjourney for visual synthesis, and Anthropic’s Claude AI models, running on retrained contextual datasets, allow these agents to reflect socio-cultural nuance and real-world variables like weather or region-specific events.
The Business Model Transformation: From Bookings to Experience-as-a-Service
Underneath the consumer appeal lies a deeper economic transformation. As AI automates booking processes and recommendation engines, companies like Expedia move toward a “platform-as-agent” economy. Instead of generating revenue per booking transaction alone, they monetize contextual relevance, partnerships, and predictive analytics delivered through their AI ecosystems.
Expedia Group CEO Peter Kern noted during a January 2025 investor call published by CNBC Markets that over 18% of their revenue came from “non-traditional” channels, including AI-personalized affiliate content. Through AI-native interfaces, brands can embed personalized modules into blogs, social posts, even WhatsApp messages. Contextual booking links and affiliate attribution convert passive content into active revenue tools.
Furthermore, predictive AI can recommend unique upselling opportunities—from hotel room upgrades to guided tours—based on a user’s behavior before finalizing the trip. Expedia now employs conversational agents not just for chat support, but as proactive sales companions that adapt dynamically with the booking funnel.
Privacy Challenges and Ethical Complexities
While excitement runs high, privacy concerns loom large. To interpret and recontextualize social media data, these AI agents need access to geolocation info, EXIF camera data, sentiment from captions, and even biometric cues (e.g., gaze or smile detection). This raises alarms around data consent, especially as boundaries blur between personal inspiration and commercial intent.
As per the FTC’s March 2025 compliance guide, companies using deeply personalized algorithms for commercial planning must ensure opt-in transparency and traceable data pipelines. Expedia promises all AI agents deployed are “transparent by design,” yet critics argue that implicit data scraping for inference—without explicit user consent—might cross ethical lines in AI marketing practice.
Implications for the Future of Travel and AI Markets
The short-term implication of these agents is convenience. The long-term implications span far wider—from redefining AI autonomy in e-commerce to expanding the role of machine cognition in lifestyle decision-making. The travel sector, long immune to deep AI automation due to its chaotic, preference-heavy nature, is now embracing LLMs as pseudo-agents that act on behalf of humans.
The ability to “reimagine intent” from ambient digital behavior, such as watching clips or liking vacation reels, makes AI agents a new limb in the planning cycle. The recent 2025 McKinsey Report on AI integration estimates that autonomous software agents—not users—will initiate over 40% of tourism discovery journeys within the next 18 months. This aligns travel planning with AI-led trends in other sectors like wealth management, HR, and education, where LLMs proactively assist rather than passively answer.
Conclusion: A Paradigm Shift in Exploration and Experience
What Expedia and Kayak are pioneering today signals a merger between exploration and execution. For decades, dreaming of destinations and booking them were disconnected. Now, the journey from discovery to decision is a single interaction. By turning social media feeds into semi-structured data inputs, processed by multimodal AI agents, users experience a new form of digital agency—one that is reactive, responsive, and rationalized in real-world logistics.
The democratization of travel inspiration, coupled with the personalization of AI itinerary generation, represents more than techno-optimism—it’s an operational blueprint for the next wave of intelligent consumer services. It’s showing the world how AI interprets not just words, but moments.