Navana.ai announced a Rs 40 crore Series A — roughly $4.2 million — on 7 September, led by entrepreneur and investor Ronnie Screwvala. Antler India, Sharad Sanghi, Sandeep Singhal and Paula Mariwala also participated. Founded by brothers Raoul and Jai Nanavati, the company sells voice AI into Indian banking, financial services and insurance.
The technical position
Navana's differentiator is deployment shape rather than raw capability: it runs on-premise for customers who cannot send call audio to a third-party cloud. Its models cover a claimed 12 Indian languages and 45 dialects, and are built for background noise, overlapping speech, code-switching and regional accents — the conditions that break general speech recognition on Indian telephony. The company says it has worked with IISc Bengaluru, IIT Madras, Microsoft Research and the Gates Foundation.
What the framing gets wrong
The claim circulating hardest is that Navana's Bodhi model achieved the highest accuracy across Indian voice AI platforms. That benchmarking exercise is described by Navana, with no published methodology, no named comparison set and no independent administrator. It is a vendor claim, not a measured result, and should be attributed as one. The same applies to the 100 million-plus voice minutes processed and the client list — Bajaj Finserv, Protean, Ujjivan Small Finance Bank, Jana Small Finance Bank, Grihum Housing Finance — all of which are company-stated.
The number that is absent
No valuation is disclosed, pre- or post-money, and none should be inferred. At $4.2m this is a small Series A by 2026 standards, and the round's significance is the thesis rather than the size: that Indian-language voice at regulated institutions is a defensible niche precisely because compliance rules out the cloud incumbents. Proceeds go to BFSI deployments, model work on languages and dialects, and an AI contact centre and evaluation platform.