Air India expands Salesforce AI for faster customer-service resolution
Air India is extending Salesforce Agentforce across email resolution, name corrections and staff knowledge support. The airline says refund turnaround has fallen from about 14 days to four hours, while name-correction processing has dropped from roughly three days to 30 minutes.
What happened
Air India is expanding Salesforce Agentforce across email resolution, passenger name corrections and staff knowledge support, aiming to automate complex service
Key facts
- Refund turnaround reduced from approximately 14 days to about 4 hours
- Name-correction processing reduced from approximately 3 days to 30 minutes
- Automated responses require a confidence threshold above 95% for eligible cases
- Digital ecosystem spans more than 140 enterprise systems
- More than 30 AI initiatives underway
Why this matters
Salesforce’s deeper integration into Air India highlights continued demand for AI platforms that combine workflow automation, agent assistance and governance, strengthening the strategic value of verticalized customer-service partnerships.
What to watch
- Reported changes in direct-booking share, app usage, Net Promoter Score and repeat-purchase rates following the rollout.
- Whether four-hour refund performance holds across peak disruption periods and agency-originated bookings.
- Expansion of AI authority from drafting and routing into executing refunds, reissues, waivers and compensation decisions.
- Rates of AI-related service errors, repeat contacts, fraud attempts and human escalations.
- New Salesforce, passenger-service-system, payment or OTA integration announcements.
- Competitor airline deployments of autonomous customer-service agents and resulting service-level benchmarks.
- Expand Agentforce into disruption rebooking, baggage claims, loyalty-account servicing and proactive flight-status communications.
- Connect AI workflows to payment, refund, passenger-service-system and OTA interfaces so cases can be resolved rather than merely triaged faster.
- Publish service-level metrics by request type, including first-contact resolution, exception rates, customer satisfaction and AI-to-human handoff rates.
- Use service-resolution data to identify recurring booking, fare-rule and digital-journey defects that generate avoidable contacts.
- Redesign staffing toward escalation specialists, quality assurance and AI knowledge governance rather than routine email processing.