Skip to content

Bolna’s inaugural Symphony brings India’s Voice AI ecosystem together, drawing 1,800+ participants in Bengaluru

India’s flagship Voice AI summit featured 42+ speakers across three stages and saw Bolna launch Syl, its in-house speech recognition model.

Maitreya Wagh & Prateek Sachan at Symphony
Maitreya Wagh, Co-founder and CEO, and Prateek Sachan, Co-founder and CTO, Bolna, delivering the keynote session at Symphony in Bengaluru, IN

Bengaluru, 20 November 2026: Bolna, the Y Combinator-backed Voice AI platform, hosted the first edition of Symphony at The Lalit Ashok, Bengaluru, with more than 1,800 attendees, including founders, enterprise leaders, investors, researchers and developers. Its 42+ speakers included leaders from Google DeepMind, General Catalyst, the Gates Foundation, NVIDIA, Sarvam AI, Exotel, Deepgram, Truecaller for Business and AT&T.

Symphony was designed as a common platform for an ecosystem that spans research, model development, speech infrastructure, telephony and enterprise applications. The one-day programme ran across three stages: Sur for keynotes and panels, Abhyas for technical builds and the Buildathon finale, and Mehfil for partner showcases and live demos.

“Voice AI is at a stage where the technology is advancing rapidly, but its adoption will depend on how well the ecosystem comes together to solve real-world challenges. We started Symphony because we wanted to bring the entire value chain into one room, from researchers and model builders to enterprises putting these systems into production. Seeing more than 1,800 people participate in our first edition is encouraging. We hope Symphony becomes a platform where the industry can exchange ideas, build partnerships and collectively push the boundaries of what voice can do,” said Maitreya Wagh, Co-founder and CEO, Bolna.

Prateek Sachan, Co-founder and CTO, Bolna, said, “Building Voice AI for India requires solving for a combination of languages, accents, latency, network conditions and cost. No single model can address every enterprise requirement, and the strongest systems will bring together different models and infrastructure. Symphony gave builders an opportunity to see these developments first-hand and understand how the underlying technology is evolving. Syl, our in-house ASR model launched at Symphony, is one step in that journey.”

Bolna also launched Syl, its in-house automatic speech recognition (ASR) model, during the Bolna Launches session, “We hear. We remember.”, presented by Bolna’s Siddharth Batra, Head of Product, and Prabal Rastogi, Head of Engineering. Syl is built for the shortest replies on a call, such as “haan”, “ji” and “theek hai”, which are easily lost on noisy phone lines and in calls that switch languages. More than 4 million calls a month on Bolna include short replies, and about one in ten of them previously came back empty. On held-out production calls, Syl transcribed 93% of one-word yes/no replies correctly, with a median latency of 250 milliseconds, at 10x lower cost than current state-of-the-art models. Deployed on an agent it had never been trained on, it cut the number of times the agent asked callers to repeat themselves by 31%, and the number of times callers repeated themselves by 47%.

In the next phase, Bolna plans to make Syl context-aware and use-case-aware and to extend it to more languages. It also previewed memory, which carries a caller’s name, preferred language, best time to call and earlier progress into the next conversation; 48% of calls on Bolna go to someone the same agent has spoken to before. Syl is the first in a planned family of Bolna models spanning ASR, LLM, text-to-speech and decision-making, with models for understanding and speaking expected by the end of the year.

In the fireside “Frontier Labs to a Billion Voices”, Dr. Manish Gupta, Vice President at Google DeepMind, discussed Project Vaani, an IISc and ARTPARK initiative that has collected more than 31,000 hours of speech across 165 districts and 105 languages, and the Vaani Atypical Speech Corpus, built with Google’s Project Euphonia to improve recognition for people with atypical speech. The session included a demonstration of a pipeline that speeds up support for underrepresented dialects.

In the same session, Srinivas Narayanan, former CTO of B2B Applications at OpenAI, in conversation with Aarthi Ramamurthy, Founder of Schema Ventures, named organisational context, security and verifiability as the main barriers to enterprise AI adoption, and argued that businesses should redesign workflows and build domain-specific evaluations rather than automate isolated tasks.

Other panels on Sur covered speech models built for human nuance, the infrastructure stack behind India’s voice agents and scaling Voice AI in Indian enterprises. A panel on Voice AI for public services brought together Sanjay Jain, Director of Digital Public Infrastructure at the Gates Foundation, Vivek Raghavan, Co-founder of Sarvam AI, Aniket Doegar of Haqdarshak and Santosh Kevlani of EkStep Foundation. On Abhyas, Exotel, Sarvam AI, Kalpa Labs and NVIDIA led masterclasses and sessions for developers.

An investor panel, “Investing in Voice AI: Who Adds Value, and What Gets Funded Next in India”, featured Akarsh Shrivastava, Partner at General Catalyst, and Pratyush Choudhury, Co-founder of Activate. The session was moderated by Thiyagarajan Maruthavanan (Rajan), Co-founder and CEO of Kalmantic Lab, and focused on how Indian startups can build globally competitive AI businesses by staying close to customers and turning India’s constraints on language, cost and latency into an advantage.

Demos went beyond customer service to FMCG retailer restocking, consumer feedback and student engagement. Voice Tank let startups pitch to investors through their own voice agents built on Bolna, followed by live questions.

The Bolna Buildathon, organised with Plivo and OpenAI, held its finals on Abhyas, where the winning team earned an interview with Grey Baker, General Partner at Y Combinator.

The Symphony Awards closed the evening on Sur, recognising Voice AI deployments in six categories: Enterprise Deployment of the Year, Voice AI for Social Impact, Bolna Partner of the Year, and Best Voice AI Deployment in BFSI, Education and e-Commerce. The Buildathon winners were announced alongside the awards.

About Bolna

Bolna is a Voice AI orchestration platform that enables enterprises to build, test, deploy and monitor human-like, multilingual AI agents at scale. Supporting 22+ Indian vernacular languages, Bolna handles 1.5M+ daily calls for customers across e-commerce, logistics, BFSI, recruitment and education sectors. The company is backed by General Catalyst, Y Combinator, Blume Ventures, and operates primarily in India while serving global customers.

Website: https://www.bolna.ai
X: https://x.com/bolna_dev
LinkedIn: https://in.linkedin.com/company/bolna-ai

Media contact:

Sonam Bala
Lead, Brand & Comms., Bolna
Email: sonam@bolna.ai
Phone: +91 91232 92954

All stories