Salesforce adds Indian languages to voice AI as enterprises move beyond pilots
Salesforce is expanding its Agentforce Voice platform to support seven additional Indian languages, betting that local language support will push more companies to move AI projects from experimental phases into production.
The company launched Hindi support for the voice platform earlier this month. Tamil, Marathi, and Telugu are currently in beta testing and will roll out in the coming weeks, according to Arun Kumar Parameswaran, EVP and Managing Director for Sales and Distribution at Salesforce South Asia. Together with English, these languages will cover 95-97% of customer language requirements across India.
The platform understands code-switching between Hindi and English, allowing customer service teams to handle mixed-language conversations without routing calls between systems.
The production problem
Salesforce sees strong demand for AI across financial services, retail, and real estate. But the company has identified a critical gap: roughly 95% of AI pilots never reach production.
"No CFO wants to put money into a technology that does not return value," Parameswaran said. Successful deployments require a clear business case and backing from senior leadership.
Indian companies are approaching AI differently than mature markets. In the U.S., customer service interactions cost $3 to $7 each, making cost reduction the primary driver for automation. In India, the same interaction costs around ₹7, shifting the focus to customer experience and revenue growth rather than labor savings.
Real results in production
Financial services firms are seeing measurable outcomes. Hero FinCorp reduced two-wheeler loan approval timelines from days to under 30 seconds using automated workflows, while improving application quality.
In real estate, Tata Realty & Infrastructure deployed Agentforce across customer engagement functions. The company reduced first response times from days to eight hours, increased lead qualification rates by 30%, and improved conversion rates by 10%.
Infrastructure and data challenges
Salesforce is also expanding local data hosting capabilities in India. The company recently announced plans to bring MuleSoft on Hyperforce to India, allowing enterprises to process and store data within the country-a requirement for regulated sectors like financial services and healthcare.
Most enterprises struggle with fragmented data spread across multiple systems and locations. "The problem is not lack of data. The problem is that the data sits in silos," Parameswaran said. Companies want to use AI but need trusted, unified data across systems to make it work.
Salesforce works with customers managing databases containing hundreds of millions of records. "If we can make technology work at India scale and cost structures, it can work anywhere else in the world," Parameswaran said.
Obstacles remain
Organizations continue to face challenges around governance, workforce skills, and employee retention during large technology transitions. The most successful deployments have clear support from top leadership. Without it, projects typically stall as pilots without measurable business outcomes.
For more on AI for Customer Support and AI Agents & Automation, explore current industry practices and tools.
Your membership also unlocks:
