The travel, tourism and hospitality industry is riding a wave of innovation driven by recent advances in agentic and multimodal AI. From synthetic query datasets for more nuanced recommendations to the rise of AI-enabled conversational agents, new research and commercial tools are enabling significant shifts in how travellers are discovered, served, and delighted. In this post, we examine the freshest developments, real measurable case studies, and what industry players need to know to stay ahead.
TL;DR
- Multimodal travel assistants like TraveLLaMA are improving map & image understanding by 6–9%, raising the bar for contextual travel help. Source
- About 31% of UK travellers used AI to book summer 2025 vacations, up 67% from 2024. AI search tools and personalisation are seen by over half of UK hospitality businesses as transformational. Source
- The global AI in hospitality & tourism market is projected to grow from US$15.69 billion in 2024 to US$20.39 billion in 2025. Source
- Agentic AI adoption is accelerating: more travel companies report cost savings, faster decision-making, better customer personalisation, and higher employee productivity. Source
- New frameworks like SynthTRIPs are helping recommender systems capture sustainability filters and under-represented user personas via synthetic queries. Source
Emerging Trend: Multimodal Intelligence for Travel Assistance
A recent academic preprint introduced TraveLLaMA, a model blending text, vision-language and map understanding to serve travellers better. It draws on a dataset combining human-forum Q&A with 90,000 image-scene/map question-answer pairs and achieved improvements of 6.5–9.4% over general LLMs on travel-specific tasks. Source
This approach matters because many traveller queries are not just textual (“Where should I stay?”) but visual or spatial (“Which way is the museum?”; “How to see nearby transit stops on a map?”). Multimodal systems which understand imagery and maps are increasingly essential for on-the-ground guidance, accessibility, and augmented reality experiences.
Rise in UK Traveller Use of AI for Booking & Personalisation
Adyen’s 2025 Hospitality & Travel Report shows that 31% of UK travellers used AI for booking summer 2025 vacations, a jump of 67% year-on-year. Younger travellers (Gen Z, Millennials) remain strong adopters, but the biggest growth came from Baby Boomers and Gen X. Source
From the provider side, 51% of UK hospitality businesses believe AI-powered search tools and AI-driven personalisation will reshape the industry. This signals a tipping point in demand—travellers expect more intelligent interactions; businesses must deliver or risk being left behind. Source
Market Size & Growth Forecasts
The “AI in Hospitality & Tourism” market was valued at approximately US$15.69 billion in 2024 and is projected to reach US$20.39 billion in 2025, growing at ~30% CAGR. Source
By 2029 the market may surpass US$58 billion, as AI tools become more embedded in operations, customer experience, and backend systems. Source
Agentic AI & Executive Insights
A McKinsey report titled “Remapping travel with agentic AI” describes how adoption is moving from experimentation to impact. Companies report measurable improvements: cost reductions, improved decision-making speed, higher quality outputs, better personalisation, and increased employee productivity. Source
Specifically, in a survey of 86 travel executives, about 26% said AI introduction has already reduced operating costs, 30% said it enables faster decision-making, 33% noted improved customer personalisation. Employee productivity gains were cited by 59%. Source
Personalised Recommendation Systems & Synthetic Query Frameworks
Existing travel recommender systems often suffer from biases, incomplete data, or underrepresentation of certain traveller needs. The SynthTRIPs framework (recent research) uses Large Language Models to generate synthetic travel queries grounded in real data, incorporating persona-based preferences (budget, style) plus sustainability filters such as walkability and air quality. Source
This enables more robust training data for recommender systems, better serving travellers seeking less common choices (off-peak, sustainable destinations, specific needs) and helping businesses anticipate demand from niche segments.
Conversational Agents: From Research to Deployment
Recent work in Artificial Intelligence-Enabled Conversational Agents in Tourism and Hospitality shows that such agents are moving beyond FAQs to nuanced customer interactions, handling complex multi-turn dialogues, resolving bookings, and even providing sentiment-aware responses. Source
These agents increasingly integrate with real-time data (weather, transport schedules, local events) and are tuned to brand voice, enabling consistent, helpful support 24/7. Positive outcomes include reduced customer service workloads, fewer escalations, and improved customer satisfaction.
Predictive Analytics & Strategic Content Planning
Tools that forecast where traveller demand will emerge allow businesses to get ahead. Smartvel describes how predictive analysis is helping airlines, OTAs and destination marketing organisations spot emerging destinations and content trends just before interest peaks. Source
By tracking indicators such as event announcements, early search volume, social media mentions, businesses can prepare content, adjust rates, schedule transport or staffing, and create packages ahead of competitors.
Case Studies: Measurable Outcomes
Case Study 1: TraveLLaMA’s Multimodal Boost
TraveLLaMA, with its new dataset combining vision, maps and text, achieved 6.5–9.4% performance improvements over general-purpose LLMs on travel understanding and visual Q&A tasks. Source
Case Study 2: UK Travellers & AI Booking Uplift
From Adyen’s findings: UK traveller AI-booking increased by ~67% from 2024 to 2025. Also, the proportion of UK hospitality businesses expecting AI search tools and personalisation to reshape the industry is over 50%. Source
Case Study 3: Global Market Growth Forecasts
The global AI in hospitality & tourism market’s growth from US$15.69 billion to US$20.39 billion within one year, and projections toward over US$58 billion by 2029, show clear economic opportunity. Source
Challenges & Ethical Considerations
- Bias & Inclusivity: Synthetic query systems must avoid reinforcing bias (e.g. favouring certain types of travellers, destinations, or budgets). Diversity in training data and representation of marginalised or under-served travellers is crucial.
- Data Privacy & Consent: Conversational agents and predictive analytics handle personal data. Transparency over data use, strong consent mechanisms, and compliance with GDPR or equivalent are non-negotiable.
- Accuracy & Hallucination: Especially for multimodal models and synthetic query frameworks, ensuring factual correctness (operating hours, public transport, safety) is vital. Grounding models with real data prevents misinformation.
- Operational Integration: AI tools must integrate with legacy systems (PMS, CRM, booking engines). Poor integration can lead to friction, frustrated users, or data silos.
- Maintaining the Human Touch: Automation helps efficiency, but travellers still value personalised human interaction. Over-automation risks losing warmth and service differentiation.
FAQ
- What is agentic AI in travel?
Agentic AI refers to systems that act with autonomy to carry out tasks (such as itinerary planning or customer support) rather than only responding to prompts. It means tools that can reason through constraints, optimise, anticipate needs, and take actions. - How much has traveller behaviour shifted in using AI?
In the UK, usage of AI for booking vacations rose to 31% in summer 2025, up 67% from 2024. - Do travellers prefer recommendations or structured bookings via AI?
Both. Many appreciate AI for inspiration, discovering activities or destinations. Others want structured, bookable itineraries and tools that integrate bookings directly. - What are synthetic query frameworks and why do they matter?
These are methods to generate realistic, diverse user queries (with filters like sustainability, budget etc.) for training recommendation systems. They matter because they help systems serve under-represented preferences and reduce bias. - How do conversational agents improve customer experience?
They offer 24/7 availability, resolve routine issues, integrate real-time data (weather, events), deliver sentiment-sensitive responses, and reduce workload on human staff. - How reliable are multimodal travel assistants?
They are improving. TraveLLaMA shows measurable gains over general models in image/map comprehension and contextual advice. - What about sustainability and ethical filters?
Increasingly built into recommendation systems and synthetic query datasets. Sustainability filters such as walkability and air quality are examples used in recent research. - Will AI replace human customer service?
No. AI handles scale, routine tasks, and provides efficiencies. Humans remain essential for empathy, problem resolution, brand character, and handling edge cases or complex issues. - What metrics matter for measuring success?
Key metrics: booking conversion rates, engagement, cost savings, customer satisfaction or Net Promoter Score (NPS), complaint or escalation rates, employee productivity, revenue per available room or ancillary income. - How soon should businesses plan to invest in these trends?
Many companies are already doing so. Given current behaviour shifts, delays risk falling behind. Investment in data infrastructure, AI tools, and ethical systems is now time-sensitive.
Mini-Glossary
| Term | Definition |
|---|---|
| Agentic AI | Autonomous or semi-autonomous AI systems that not only respond but plan, predict, and take action under constraints. |
| Multimodal Model | An AI model that handles more than one form of input—text, images, maps, sometimes audio or video—to understand richer context. |
| Synthetic Query | A generated prompt or question used for training recommendation systems to cover diverse preferences, constraints, or under-served user groups. |
| Recommender System (TRS) | Algorithms or frameworks that suggest content, activities, destinations or services to users based on preferences and data. |
| Personalisation | Tailoring offers, experiences, content, or suggestions to individual user preferences or history. |
| Conversational Agent | An AI system designed for human-like dialogue, assisting with queries, requests, bookings, etc. |
| Predictive Analytics | Using historical and real-time data to forecast trends, demand, or interest so decisions can be proactive. |
| Sentiment Analysis | Assessing emotion or opinion in text (reviews, chat logs etc.) to adjust responses or services accordingly. |
| CAGR (Compound Annual Growth Rate) | The rate at which a market grows year over year, compounded—used in forecasting market size. |
| LLM (Large Language Model) | A machine-learning model trained on massive datasets to generate or understand human-like text. |
| Data Grounding | Ensuring AI outputs are based on verified, real-world data to avoid hallucination or misinformation. |
| Sustainability Filter | Criteria (like walkability, air quality, environmental impact) included in recommender systems or queries that help travellers prioritise eco-friendly options. |
| Accessibility | Designing travel services or AI tools to be usable for people of varying physical abilities, sensory needs or mobility constraints. |
Conclusion & Takeaway
The latest advances in agentic and multimodal AI, synthetic query frameworks, and rapidly rising adoption among travellers and hospitality businesses are reshaping how the industry operates. Success now hinges not simply on having AI, but on making it contextually smarter, ethically grounded, and tightly woven into customer journeys.
Comments
Leave a Comment
No comments yet. Be the first to comment!