ElevenLabs plans to expand voice AI services to all 22 official languages of India

CEO and co-founder of ElevenLabs, Mati Staniszewski, stated that the success of artificial intelligence (AI) implementation will depend entirely on how well the technology interacts with people. He emphasized that in a significant part of the economy, interpersonal skills are as important as raw computational power.

Staniszewski noted that creating an impressive experience requires a combination of both factors, and people must be able to interact with AI as naturally as they interact with each other—through conversation and voice, using their language and dialect. These remarks were made at the first ElevenLabs Summit in India, held in Bengaluru on Tuesday.

The London-based company already provides speech-to-text and text-to-speech services in 14 Indian languages and intends to cover all 22 official languages of the country within the next year. This is driven by a desire to deepen its presence in a market the company considers its largest outside the US.

Staniszewski added that if the challenge of human-AI interaction can be solved in India, this solution can also be applied anywhere else in the world. The company's goal is to create AI that is natural, intuitive, and trustworthy in daily conversation. He clarified that simply sounding human is only the first step; AI must also be able to listen, understand context, respond naturally, adhere to constraints, and act according to intent.

The field of voice AI is experiencing rapid growth, attracting significant investment from global investors. According to Tracxn, there are 153 companies in the voice AI sector globally, of which 107 have received funding, collectively raising $3.6 billion in venture capital and private equity as of August this year.

India is viewed as the largest testing ground for the potential of this sector due to its population of 1.4 billion people, hundreds of millions of users who still use feature phones, and over 100 active spoken languages. Nandan Nilekani, Chairman of Infosys and architect of Aadhaar, believes that voice AI is the last frontier and the only practical interface for achieving true digital equality in India.

ElevenLabs collaborates with over 250 enterprises and companies across India, including Razorpay, TVS Motor, and Urban Company. In the last year, Indian companies conducted 100 million dialogues through ElevenAgents, with over 70 percent of them being in Hindi, Kannada, Tamil, and Telugu.

The company has achieved a global revenue run rate exceeding $600 million and plans to hire 100 employees in India. Furthermore, ElevenLabs has partnered with the Government of Karnataka to explore the implementation of voice AI in Kannada in areas such as upskilling, public services, and grievance redressal, while simultaneously working on AI ethics and voice restoration.

Staniszewski and Karnataka's Minister of IT and Biotechnology, Priyank Kharge, officially formalized the company's first partnership with the state government of India through a Memorandum of Understanding with the Karnataka Innovation and Technology Society (KITS). This partnership will focus on exploring the application of voice AI in Kannada in fields like education, citizen services, and complaint resolution, as well as working on responsible AI and voice restoration.

Similar stories

techcentral.co.zaSeptember 30, 11:53

How Voice AI is Reducing Costs and Increasing Efficiency in South African Contact Centers

The contact center has become one of the most expensive points for answering simple customer inquiries, whether it's checking order status, confirming a balance, or changing a delivery time. When an agent takes a call, finds the account, reviews the details, and concludes the conversation, it takes about two minutes of the agent's working time. Multiplying this figure by 300 calls per day results in significant costs arising from processes that do not require human involvement.

Voice artificial intelligence is changing this calculation model. It does not replace the contact center itself, but rather handles the volume of requests that should not have reached the agent in the first place. Companies successfully implementing this technology are not those that completely rebuild their infrastructure, but those that add an intelligent voice layer to existing systems and begin to record positive changes within a few weeks.

The main problem lies not with personnel, but with architecture. Contact centers in South Africa face simultaneous pressure from all sides. Call volume sharply increases at the end of the month, on weekends after payday, or when digital channels fail and customers switch to the phone. Agent capacity cannot adapt to demand, leading to increased waiting times in queues and a rise in abandonment rates. Furthermore, an agent who does answer may handle a fifteenth identical request during a shift.

Traditional contact center setups exacerbate this problem because they were not designed to handle the modern diversity and volume of customer contacts. Disparate systems for voice, WhatsApp, email, and social media force agents to constantly switch between screens, which eliminates the overall context of the interaction. Every conversation starts from scratch. Department managers compile reports from four different platforms but still cannot get a complete picture of what is happening in the queue at any given moment. This is an architectural problem, not a staffing issue, and the fastest way to solve an architectural problem is to add a connecting layer over existing systems.

Voice AI Functionality

Voice AI represents an intelligent layer positioned before telephony that processes incoming calls using natural language understanding. The customer vocalizes their need, the system interprets the intent, verifies the relevant data, and either resolves the request automatically or routes the call to the appropriate agent with fully loaded contextual information.

AI performs most effectively with predictable requests: account authentication and verification, balance and statement inquiries, tracking order and delivery status, scheduling and rescheduling appointments, payment reminders, and off-hours support across all these areas. These are precisely the interactions that consume the largest share of agent time in most South African contact centers.

However, AI is not intended to replace an agent when dealing with complex complaints, sensitive debt collection conversations, or requests requiring judgment and negotiation. It handles the high volume, allowing the agent to focus specifically on those interactions where a qualified specialist can make a measurable impact on the final outcome.

Application of Voice AI in South African Industries

In the retail and e-commerce sectors, voice AI automates transactional requests—such as order tracking, returns, and delivery updates—during peak demand periods like Black Friday and paydays. By integrating with WhatsApp and web chat, it ensures consistent responses across various channels and frees up human agents to handle complex issues.

In the financial sector and debt collection, AI operates 24/7, managing authentication, balance checks, payment reminders, and delinquency notifications. With built-in compliance with the Protection of Personal Information Act (POPIA), it automatically masks sensitive data, records and transcribes every call, ensuring 100% automated auditing for managers.

For Business Process Outsourcing (BPO) providers, voice AI boosts productivity by increasing retention rates (the percentage of calls resolved without agent intervention), reducing average handling time, and providing multilingual 24/7 campaign support without proportionally increasing staff. Integration with existing telephony systems gives BPOs a key competitive advantage.

In healthcare and public services, AI manages appointment scheduling, queue updates, prescription reminders, and answers general frequently asked questions. By processing a large influx of calls, it reduces waiting times and allows frontline staff to concentrate on critical personal patient care.

Consequences of Operating Without AI

Every month, a contact center operating without voice AI pays for agent time spent on calls that a voice agent could handle in mere seconds. It incurs costs related to staff turnover among agents fatigued by routine requests. Moreover, it loses calls received outside business hours and misses potential customers. The company also manually conducts compliance checks only on a sample of calls instead of analyzing every single one.

Companies collaborating with 1Stream in the retail, financial services, and BPO sectors consistently note that the volume of inquiries absorbed by voice AI in the first 90 days exceeds initial estimates made during the implementation phase. The reason is simple: once they start measuring exactly where agents spend their time, the proportion of transactional calls is almost always higher than initially assumed.