How HBX Group Uses Google Gemini to Automate 30% of B2B Travel Support Cases

HBX Group operates at massive scale in B2B travel, connecting 60,000 travel distributors to hotels and providers across 190 countries. Each month, the company manages more than 127,000 customer service cases — nearly 60,000 emails and 50,000 phone calls — spanning multiple languages, time zones, and levels of complexity. The support operation had to handle both highly routine requests (fee waivers, booking confirmations, hotel confirmation numbers) and complex, multi-party escalations requiring deep expertise.

The challenge was structural. A system built for this scale had to work across four channels—chat, email, helpdesk, and voice—in multiple languages, integrated with core systems including Salesforce CRM and Vonage telephony, while remaining flexible enough to evolve as AI technology advanced. Standard chatbot deployments wouldn’t meet this bar. What HBX Group needed was a genuinely omnichannel AI platform with enterprise-grade reliability.

Working with Ayesa Digital, a Google Cloud partner, HBX Group built an omnichannel contact center platform on Google Cloud Contact Center AI (CCAI) and Dialogflow CX, powered by Gemini models for natural language understanding. Implementation followed a “Quick Wins Strategy”—automating text channels first (chat, email, helpdesk) before tackling live voice, which reduced deployment risk while building a shared foundation of components reusable across channels. Gemini models enabled the system to grasp intent, sentiment, and nuance across languages, while Google’s Chirp 3 voices created a natural-sounding experience for phone interactions. ADK (Agent Development Kit) powers the ongoing transition to fully agentic AI capabilities.

The results are measurable across every channel. All client requests are now classified with a specific category and sentiment. In targeted areas—hotel confirmation numbers and fee waivers—automation handles 32.5% of cases without human involvement. The intelligent voice channel automates 18% of high-volume live phone calls. Multilingual chat operates 24/7 using real-time LLM translation. The platform has a single compounding learning loop: improvements in one channel automatically improve others.

HBX Group and Ayesa Digital are now building the next phase: a fully agentic system where AI agents can reason, plan, and execute multi-step tasks across booking systems and back-office platforms without human intervention. The roadmap targets 80% of interactions passing through AI automated resolution by late 2026.