Air India deployed an agentic AI solution (AI.g) built on Azure that now handles approximately 40,000 customer queries daily across 1,300+ question types, achieving a 97% resolution rate since launch six months prior. The airline faced the classic scaling problem: millions of inbound queries creating bottlenecks, rising operational costs, and deteriorating response times. Rather than bolting AI onto existing infrastructure, Air India's internal team built a purpose-built agentic system that could learn and improve iteratively from real interactions. The scale of deployment is notable—13 million conversations resolved—and positions Air India as the first airline to operationalise generative AI for customer service at this magnitude. This raises a critical question for CX leaders: if a legacy airline can build and deploy custom agentic solutions in-house within six months, what does this mean for the competitive positioning of established platforms like Zendesk and Freshdesk that are still integrating AI as a feature rather than a foundational architecture?
The implications for support teams are twofold. First, the 97% success rate suggests that agentic systems handling transactional queries (booking changes, refunds, status checks) are now production-ready, which means teams should be evaluating whether their current platform's AI capabilities can genuinely deflect volume or merely triage it. Second, Air India's approach—building internally rather than licensing—demonstrates that the barrier to entry for custom agentic solutions has collapsed. This creates pressure on traditional CX platforms to either deepen their AI capabilities substantially or risk becoming orchestration layers for best-of-breed agentic tools. The related acquisitions of Fin by Salesforce and emergence of platforms like ChatSpark's AI Operator suggest the market is already fragmenting around this realisation. For teams already embedded in Zendesk or Freshdesk, the question becomes whether incremental AI improvements will suffice, or whether you need to evaluate agentic-first alternatives that can match Air India's throughput and resolution rates.
How Azure AI helped Air India reinvent customer service by answering 40,000 daily queries instantly Microsoft
How Azure AI helped Air India reinvent customer service by answering 40,000 daily queries instantly microsoft.com