AI Goes Offline: Indian Startup Puts Intelligence on Feature Phones
Sarvam AI deploys edge models on Nokia feature phones and smart glasses, challenging the cloud-first AI paradigm with offline-capable intelligence for emerging markets.
While Silicon Valley obsesses over cloud-based AI requiring constant internet connectivity, an Indian startup is betting on the opposite: AI that works when you're completely offline.
Sarvam AI, backed by Lightspeed, PeakXV, and Khosla Ventures, just announced partnerships to deploy its edge AI models on Nokia feature phones, cars, and its own smart glasses. The twist? These models occupy just megabytes of space and run on existing processors without any internet connection.
The Feature Phone Revolution
At the India AI Impact Summit in New Delhi, Sarvam demonstrated something that would make most AI engineers do a double-take: a user pressing a dedicated AI button on a basic Nokia phone to converse with an assistant in local languages about government schemes and market prices.
The company is partnering with HMD Global to integrate conversational AI into Nokia and HMD devices. "Through edge AI, we want to bring intelligence to every phone, laptop, car, and even a new generation of devices," said Tushar Goswamy, Sarvam's head of Edge AI.
This isn't just about nostalgia for simpler phones. In India, feature phones still command a significant market share, particularly in rural areas where internet connectivity remains spotty and data costs matter. By putting AI directly on the device, Sarvam is potentially reaching millions who've been left out of the AI revolution.
The Sovereignty Play
Sarvam's approach aligns with India's push for "sovereign AI"—keeping data and processing within national borders. The company has partnered with Qualcomm to optimize its models for the chipmaker's processors, with Qualcomm developing a "Sovereign AI Experience Suite" for phones, PCs, cars, and IoT devices.
"This will allow Sarvam to design models and applications that run closer to the edge, safeguard data, and are ready for adoption, at scale," said CEO Vivek Raghavan.
The startup is also working with German engineering giant Bosch to bring AI assistants to cars and showcased its own AI smart glasses, dubbed "Sarvam Kaze," designed and manufactured in India.
The Market Shift
This represents a significant pivot for Sarvam, which has primarily operated in the enterprise market offering voice-focused models for customer support. The move toward consumer applications suggests the company sees untapped potential in serving users who can't or won't rely on cloud-based AI.
The timing is intriguing. While competitors race to build larger models requiring more computational power and internet bandwidth, Sarvam is going the opposite direction—smaller, lighter, and completely self-contained.
This content is AI-generated based on source articles. While we strive for accuracy, errors may occur. We recommend verifying with the original source.
Related Articles
Raspberry Pi launches the $130 AI HAT+ 2 for Raspberry Pi 5. Featuring 40 TOPS performance and 8GB RAM, it enables local execution of generative AI models like Llama 3.2.
Explore the Kazakhstan AI Digitalization Strategy 2026. President Tokayev's 3-year plan involves supercomputers, AI education for 650k students, and a new AI ministry.
Ugreen unveils the iDX6011 Pro local AI NAS starting at $1,699. Learn about the specs, privacy benefits, and why local AI search is commanding PC-level prices.
Google has released FunctionGemma, a 270M-parameter Small Language Model for on-device function calling. Discover how this edge AI model offers a privacy-first, low-latency, cost-effective alternative to cloud-based LLMs.
Thoughts
Share your thoughts on this article
Sign in to join the conversation