Navana.ai

We're setting up our dedicated QA function at Navana. We started as a small team with a big vision: India deserves voice AI built for how India actually speaks. And today our models run live inside banks, insurers and lenders. Real calls, real people, real money on the line. Growth has been real fast, and now the quality bar has to match it. So we're hiring the person who builds QA at Navana from zero. No inherited process. No playbook to follow. Just a team that ships fast, cares about craft, and wants someone to make quality the thing that lets us go faster. This is for the person who: - Has built quality where none existed and still thinks about what they'd do differently. - Gets a little obsessed with what breaks at 10x. - Doesn't stop at "tested," but follows a release to see whether it held true. - Would rather draw the lanes themselves than inherit them. You don't need voice AI domain experience nor a big-company title. You need the instinct, the craft, and the appetite to build and grow with us. As the founding QA hire, you shape the process, the next 10 hires & everything after. If this sounds like you, write to joinus@navanatech.in with your resume and one thing you built to make a product more reliable. JD here: https://lnkd.in/g3Jc8ReD

We're setting up our dedicated QA function at Navana. We started as a small team with a big vision: India deserves voice AI built for how India actually speaks. And today our models run live inside banks, insurers and lenders. Real calls, real people, real money on the line. Growth has been real fast, and now the quality bar has to match it. So we're hiring the person who builds QA at Navana from zero. No inherited process. No playbook to follow. Just a team that ships fast, cares about craft, and wants someone to make quality the thing that lets us go faster. This is for the person who: - Has built quality where none existed and still thinks about what they'd do differently. - Gets a little obsessed with what breaks at 10x. - Doesn't stop at "tested," but follows a release to see whether it held true. - Would rather draw the lanes themselves than inherit them. You don't need voice AI domain experience nor a big-company title. You need the instinct, the craft, and the appetite to build and grow with us. As the founding QA hire, you shape the process, the next 10 hires & everything after. If this sounds like you, write to joinus@navanatech.in with your resume and one thing you built to make a product more reliable. JD here: https://lnkd.in/g3Jc8ReD

Our Co-founder and CEO, Raoul Nanavati, was recently featured in The CapTable, weighing in on a question India’s AI ecosystem will soon have to answer: where should it actually compete? The answer isn’t necessarily to build the biggest model. For enterprise AI, smaller, more specialised models often mean better accuracy, more control, and lower inference costs, especially when they are built for a specific language, industry, or workflow. This has always been at the heart of what we believe when serving enterprises and institutions. It needs to work inside real processes, with control over models, data, deployment, and evaluation. That's how we build at Navana. Our proprietary speech models are designed from the ground up for Indian speech, built for large-scale deployment, and rigorously benchmarked and evaluated. Thank you to The CapTable for spotlighting where India’s AI ecosystem can build its advantage. Read the full article here: https://lnkd.in/dx37Eh8c #AI #EnterpriseAI #VoiceAI #NavanaAI

Bodhi TTS (Text-to-Speech) was recently featured by Economic Times CIO! We’ve always believed that the larger shift underway in enterprise Voice AI for India was never about whether an agent can simply sound human, but whether it could work for India: in its languages, dialects, workflows, and real customer conversations unique to our country. With Bodhi TTS, we now aim to extend the same India-first approach that powers our deployments into the speech generation layer that supports our largest institutions. Deeply grateful to ETCIO for bringing attention to an issue we’ve been hard at work trying to solve! Sign up now at: http://platform.navana.ai The first 100 approved BFSI builders receive 100 hours of credits! Read the full article here: https://lnkd.in/dgvn8H-Y #VoiceAI #BodhiTTS #EnterpriseAI #BFSI #NavanaAI

Bodhi TTS is live! We built it to behave like speech infrastructure, not a demo voice. Our voice layer is built from the ground up with India in mind. It includes 50+ natively Indian-sounding voices, zero-shot cloning from a few seconds of reference audio, consistent voice identity across 10 languages, and pronunciation control, ensuring steady updates without retraining models. Audio starts in <100ms. Our model runs at roughly 67x real-time across all 10 languages on a single L40S, and it is compact enough to run comfortably on an L4. This matters because inference cost is where production Voice AI gets expensive very quickly. Bodhi TTS is now live for early access, and the first 100 approved BFSI builders get 100 hours of credits, enough to run a real production workload end-to-end on your own audio. Sign up now: platform.navana.ai #NavanaAI #BodhiTTS #VoiceAI #SpeechAI #APIs 

The best-known TTS models keep getting bigger. We went the other way. Bodhi TTS is roughly 5x smaller than the nearest comparable model, and that one choice shapes everything else: ⚡ Under 100ms to first audio byte, so voice agents respond without awkward pauses and calls feel like conversations 💰 Runs on low-cost, widely available GPUs, which is how it's priced for high volume 🎙️ Zero-shot voice cloning: a few seconds of audio, no training, and the same voice speaks all 10 languages 🗣️ Pronunciation you can instruct: the LLM can tell it exactly how to say a customer's name, and ₹1,00,000 is read as एक लाख रुपये Small, fast, and built for how India speaks. Proud of the team that built this.

The Navana API Platform is now live! Our Bodhi TTS (Text-to-Speech) is open to everyone, offering 50+ Indian voices across 10 languages, <100ms time to first audio, zero-shot voice cloning, and custom pronunciation controls. Bodhi STT (Speech-to-Text) is available on limited access for BFSI builders, with 10 languages, 40+ dialects, native code-switching, and models tuned for real telephony and banking workflows. One platform. Both sides of the speech stack. Built for how India actually sounds. Three ways to get started with Bodhi: - BFSI early access: Full STT + TTS access, with the first 100 approved teams receiving 100 hours of usage credits, valid for 90 days. - Open sign-up: TTS is available to everyone from day one. STT remains available on request for select BFSI builders. Sign up here: https://lnkd.in/dxKRiJXN #NavanaAI #VoiceAI #BFSI #BodhiSTT #BodhiTTS

The one with the team. ✨ They say Rome wasn’t built in a day. Neither was Bodhi. A lot went into getting us to Global Fintech Fest — weeks of building, testing, fixing, preparing, rehearsing, and then showing up for three straight days to make every conversation count. The booth, the demos, the Bodhi Text-To-Speech launch, the customer conversations — none of it happens without the people behind it. Those conversations are the real test. They tell you very quickly what resonates, what doesn’t, and where the product still has work to do. GFF was great for meeting customers and opening up new opportunities, but one of the best parts was seeing the team bonding that happened as a result of it. Three intense days, hundreds of conversations, and the same energy right through to the end. Big shoutout to everyone at Navana who showed up, jumped in, and made every interaction count. #GlobalFintechFest #TeamNavana #VoiceAI #BuildingForIndia