This page collects 20 real-world use cases of AI agents in travel & hospitality. 14 are verified by a production case study or reproducible demo; 6 are unverified reports or concepts, and they are labeled as such. The most common autonomy level here is Level 3 (Conditional): autonomous within guardrails; humans handle exceptions. Business functions covered include Customer Service, Legal Ops, Software Development and Marketing. Vendors and tools mentioned include Legora, Microsoft (Azure OpenAI), Airbnb (in-house build) and Alibaba Cloud (Qwen). Each entry lists the workflow, tools, human-in-the-loop points and risks, and shows outcomes only when a source reports them, with a link to that source.
Accor's global in-house legal team of around 140 lawyers adopted Legora after a six-month expert trial, using its Word/Outlook add-ins and tabular review for drafting, review and large-scale contract portfolio analysis; Accor reports about 99% regular usage and 90%+ time reductions on some work (reported by Legora / Accor)
Air India began building AI.g on Azure OpenAI after Microsoft launched the service in November 2022, and calls it the first airline generative AI customer-service deployment at scale. Microsoft's customer story (February 11, 2026) says AI.g handles about 40,000 queries a day across more than 1,300 question types, from booking changes to refund requests. It has resolved 13 million+ conversations at a 97% success rate, so only 3% go to a human (reported by Microsoft, the vendor). In the same story, CDTO Dr. Satya Ramaswamy cites '97% of 4 million-plus customer queries', which doesn't match the 13 million figure. It is mainly an answer assistant; agents that run full workflows such as refund processing are still in trial
Automated pipeline migrated ~3,500 Enzyme test files to React Testing Library with validation and retries (reported by Airbnb)
Airbnb started its AI strategy at the bottom of the funnel with customer service. It rolled out an AI customer service agent in spring 2025 and expanded it to all US English users in May 2025. The agent is built on 13 models from OpenAI, Alibaba, Google and open-source providers, relying mostly on Qwen because it is 'fast and cheap'. It shows action buttons and links so guests can complete tasks such as reservation changes or cancellations. It must also work across languages, follow nearly 100 policies and escalate accurately to humans, especially for trust and safety incidents. In October 2025 Airbnb said the agent had cut users needing a live representative by 15% and average resolution time from nearly three hours to six seconds (reported by Fortune/Bloomberg). On the Q1 2026 earnings call CEO Brian Chesky said it resolved over 40% of inquiries, up from about a third the prior quarter. He also said cost per booking fell about 10% year over year, which can't be attributed to the agent alone (reported by CX Dive). The figures are company-reported, but they come from earnings disclosures over about a year of production
Engine (online travel for businesses and groups; 800,000+ client-service requests a year) replaced a chatbot that could not take action with 'Eva', an Agentforce agent on chat and voice. Eva uses retrieval from Salesforce Data 360 (Service Cloud history, 10M+ rows of hotel data from Snowflake and Amazon S3, Confluence guidelines) with LLMs on Amazon Bedrock, confirms who the customer is and what they may change, calls booking-platform APIs to cancel or change hotel, flight and car bookings, and hands complex group rebookings to a rep with a case and full context. Built with NeuraFlash and Salesforce Professional Services in under three months. Salesforce reports 50% of chat cases resolved automatically, a 16% chat CSAT increase and a 15% cut in average handle time (reported by Salesforce, citing Engine; vendor-published)
CBRE uses Levelpath AI-native sourcing so teams describe a buy and generate RFx packages instantly, then AI evaluates proposals in any format; in one 42-location janitorial event, RFx build took 15 minutes, bid analysis under a minute, award in one meeting, launch a month early and 9% additional savings vs a historical 6–8 week cycle (reported by Levelpath / CBRE)
Day-use hotel booking platform Dayuse has a three-person CRM team that puts BrazeAI Agent Console agent steps inside Braze Canvas journeys; each step writes copy for one user from their wish-listed hotels, booking count, property type and preferred language. Moving from a ChatGPT webhook to Agent Console gave a further 23% uplift on the repeat campaign; the earlier LLM-written wishlist campaign had doubled incremental revenue vs. a control group (reported by Braze)
Holland America Line's Anna agent on Copilot Studio helps guests and travel advisors book cruises, add services, and answer itinerary questions (reported by Microsoft / Holland America Line)
ISS North America, a facility services company whose frontline workforce works across airports, hospitals, corporate campuses and manufacturing sites, made Paradox's conversational AI 'Iris' the front door for every candidate. Iris handles screening, interview scheduling and onboarding paperwork by text message in the candidate's preferred language, with no logins, around the clock. ISS's own blog reports an 85% reduction in time-to-hire (from 65 days), a 97% application completion rate and more than 10,000 recruiter hours returned (reported by ISS North America blog, April 2026, Chief People and Culture Officer Americas Lisa Mitchell-Kastner as author). Workday's customer story adds that 35% of conversations happen after business hours (reported by Workday). Neither source gives a measurement window or names the model, and the gains come alongside a wider redesign of the application process.
UK facilities-services group Mitie rolled out SAP's Joule AI assistant, built into SAP SuccessFactors, to all employees as one of SAP's earliest customers to do so, so staff can get instant answers to everyday HR questions instead of raising queries with HR (reported by Mitie via the Shared Services Forum UK). Mitie says that since the October launch more than 6,000 employees have used Joule and they have self-served over 65,000 times (reported by Mitie; adoption figures only, no measured drop in HR queries is published in these sources). Its SAP Innovation Awards 2026 entry says Joule handled over 13,000 interactions in its first month (reported by Mitie's SAP awards entry; usage figure only). Mitie won SAP's People Experience Pioneer award for the rollout.
Two years ago a Spring upgrade took a Navan senior staff engineer over a month and a bad merge locked master for a day. For the Spring Boot 4 migration he planned the upgrade once in Cursor, then delegated each component to Cursor Cloud Agents running on Navan's self-hosted machines, where agents use internal tooling and MCPs for Snowflake and Slack and spin up browsers to verify their own changes. Eight applications and 15 libraries moved at the same time while he monitored the agents, and the migration finished in about a week without freezing development. All generated code goes through the usual human and AI review, testing and staging. About 200 Navan staff now actively use Cloud Agents (reported by Cursor, citing Navan; vendor-published, single migration)
OFS Group (founded 1966, 50-100 staff) had aging receivables after growth and a systems transition. Abivo's voice agent Kate calls customers about outstanding invoices using balance, due-date and history data, holds payment conversations, clarifies disputes and timing objections, and captures promises to pay and follow-up needs, syncing a summary to Abivo's dashboard for the OFS team. Abivo reports $842,518 in overdue payments recovered within four months of go-live, DSO down by more than 30 days, and the collections team going from four callers to one (reported by Abivo, quoting OFS Group's CEO and Financial Controller; vendor-published)
Road Scholar partnered with Nansen to deploy Optimizely's Agent Platform (Opal) across marketing and program operations, reporting ~92% faster content creation, about 2 hours saved per operational task, ~60% Opal adoption, and a headline 98.5% reduction in operational workflows (reported by Nansen / Road Scholar)
Ryanair, Europe's largest airline, replaced a static legacy chatbot with an AI customer-service agent built on Amazon Bedrock AgentCore and Amazon Nova 2 Lite with AWS Partner Cation Consulting. It covers baggage, refunds, bookings and flight disruptions. AWS's case study reports 120,000 chats a day in seven languages at an 80% containment rate, and 10 million answers at 94% accuracy since the October 2024 launch (reported by AWS, the vendor; PYMNTS re-reported the case study and added no independent data). Response time fell from 18 to 2.9 seconds after a five-model evaluation, and twelve domain agents were merged into one in February 2026. The 500,000+ agent hours saved and the 70% cut in agents per passenger combine this chat agent with Amazon Connect voice work, so they are not chat-only results
Adeera, a PIF group hotel brand operator with about 5,000 staff, faced growing contract volumes with a short-staffed legal department and reviews taking 2–4 hours each. It adopted LEGALFLY for clause review, contract translation, automated reporting and expiry-date tracking; the tool adapts to the General Counsel's preferred wording, for example defaulting to his revised insurance clause after a few edits, and drafts go through peer review. LEGALFLY reports review time cut from 2–4 hours to 10–15 minutes per contract (93%), more contracts reviewed with the same team, and more consistent clauses (reported by LEGALFLY; General Counsel Dries Deforche quoted).
Paradox conversational agent handles application and scheduling for high-volume hiring.
Southwest Airlines, whose 2,600+ service reps handle more than 20 million customer contacts, began a phased rollout of Agentforce in November 2025, its first agentic AI rollout. The agent answers only from Southwest's approved knowledge base and Service Cloud data. For authenticated tasks such as reservation changes it securely deep-links customers to the booking website rather than making the changes itself. When it can't resolve an issue it carries the conversation to a rep through Enhanced Chat or helps request a callback. Salesforce's story (October 13, 2026) reports a 45% case resolution rate across 2 million+ interactions. It also gives $6 million in projected annual savings and a 7x ROI, both estimates rather than measured savings (reported by Salesforce, the vendor)
Business-travel platform TravelPerk's legal team (GC Andrew Cooke) deployed Flank in May 2023 as an #ask-legal Slack agent that grew to cover privacy, security, revenue and bids, plus a Virtual Deal Coach for sales negotiations. It handles about 3,000 queries a month (peak 5,000), automatically services ~1,200 privacy queries a month and cut privacy questionnaire completion from weeks to hours, so sales and deal desk answer RFP and vendor forms on their own (reported by Flank)
trivago's hotel-search web frontend organization (seven feature teams, ~30-35 developers) adopted CodeRabbit to make reviews consistently thorough; PR authors must address CodeRabbit or GitHub Copilot findings before requesting human review, so people focus on architecture and product logic. CodeRabbit caught a state-management error humans had missed and a local-storage eviction bug plus its missing test, and an ignored browser-compatibility warning was later linked to an estimated 210,000 broken sessions and up to €100,000 potential impact. In a 90-day snapshot it supported 1,160 merged PRs, developers accepted 43.4% of 1,620 comments, and it saved an estimated 2.07 weeks of reviewer time (reported by CodeRabbit, citing trivago)
Xanterra Travel Collection deployed Cresta AI Agents (Faith, Cali, Skye) for national-park hospitality guest services at ~74% average containment; separately, Cresta reports a $3.3M revenue lift for its whole platform (including Agent Assist and Analyst), not the AI agents alone (reported by Cresta)