Industry reports indicate that there’s an increasing demand for voice AI solutions in India, as both individuals and businesses lean towards this communication medium. This trend has sparked interest among enterprises and startups eager to harness voice AI for improved customer support, sales, acquisition, hiring, and training.
However, recognizing market demand is one thing; demonstrating that businesses are willing to pay is another matter altogether. Y Combinator, the prominent startup accelerator, turned down Bolna — a voice orchestration startup founded by Maitreya Wagh and Prateek Sachan — five times before finally accepting them into the fall 2025 batch. Their skepticism stemmed from doubts about whether the founders could translate interest into revenue.
“When we applied to Y Combinator, the feedback we received was, ‘Great product that creates realistic voice agents, but Indian companies aren’t going to pay, and you won’t make money from this,’” Wagh shared with TechCrunch.
The startup reapplied with the same concept for the fall batch but showcased that it had generated over $25,000 in monthly revenue for several months. Initially, Bolna was running $100 pilots to help users build voice agents, but those pilots are now priced at $500 each.
Momentum has been building since then. On Tuesday, Bolna announced it raised a $6.3 million seed round led by General Catalyst, with contributions from Y Combinator, Blume Ventures, Orange Collective, Pioneer Fund, Transpose Capital, and Eight Capital. The funding round also features individual investors like Aarthi Ramamurthy, Arpan Sheth, Sriwatsan Krishnan, Ravi Iyer, and Taro Fukuyama.
Bolna aims to create an orchestration layer — a platform that connects and manages various AI voice technologies — similar to startups like Vapi, LiveKit, and VoiceRun, tailored for the unique interaction style in India. This includes features like noise cancellation, integration with Truecaller for verification, and support for multiple languages.
The company has introduced specific functions for Indian users, such as speaking numbers in English regardless of the primary language spoken or permitting keypad input for longer responses.
Wagh emphasized that Bolna’s key differentiator is its user-friendly approach, allowing users to create voice agents by simply describing them, even if they lack technical expertise. As a result, 75% of the company’s revenue comes from self-serve customers.
He mentioned that, as an orchestration layer, Bolna isn’t tied to a single model. This flexibility allows enterprises to switch to a better model when needed. “Our platform enables customers to easily change models or even use different ones for specific locales to maximize effectiveness. An orchestration layer is essential for businesses to always utilize the best models, as the top-performing model can change over time,” Wagh explained.
Bolna caters to a diverse clientele, including the car reselling platform Spinny, the on-demand house-help startup Snabbit, beverage firms, and dating apps. Most clients are small to midsize businesses leveraging Bolna’s self-service platform.
Additionally, Bolna is targeting larger enterprise contracts. For these, they have a team of forward-deployed engineers who work directly with clients, either onsite or closely allied with their teams. So far, Bolna has secured two large enterprises as clients and has four more in the pilot phase. Currently, Bolna employs nine forward-deployed engineers and plans to add two to three more each month to support this enterprise endeavor.
Bolna has recorded steady growth in both call volumes and revenue. They are now processing over 200,000 calls daily and are nearing $700,000 in annual recurring revenue (ARR). The company noted that while 60% to 70% of call volume is in English or Hindi, the use of regional languages is on the rise.
Akarsh Shrivastava from General Catalyst stated that the firm found Bolna compelling because its orchestration layer offers flexibility for various customer needs. “Bolna provides the freedom to choose any model and has the infrastructure to adapt it to your requirements. It’s an excellent choice for those who want to have partial control over the stack, desire flexibility in model selection, and wish to maintain those products themselves,” Shrivastava told TechCrunch during a call.
