Shiprocket launches indigenous multimodal AI model Shunya.ai for small businesses

Shiprocket launches indigenous multimodal AI model Shunya.ai for small businesses

Shiprocket has launched Shunya.ai, a sovereign multimodal AI model, as the IPO-bound logistics platform looks to embed deeper into the workflows of small and mid-sized businesses.

Built in partnership with US-based Ultrasafe Inc., Shunya.ai is trained on Indian commerce data, supports nine Indian languages, and is hosted entirely within India. The model is being pitched as a locally compliant alternative to global large language models (LLMs), with a specific focus on micro, small and medium enterprises (MSMEs) and direct-to-consumer (D2C) sellers.

The launch comes amid heightened interest in India’s AI ecosystem, with growing regulatory push for data localisation and domestic compute infrastructure.

The backend for Shunya.ai will run on GPU infrastructure from L&T’s Cloudfiniti, which, Shiprocket says, ensures compliance with India’s digital sovereignty norms.

“We’ve built Shunya in partnership with Ultrasafe using their proven LLM infrastructure, and layered it with our own data and distribution,” Saahil Goel, MD and CEO of Shiprocket, told Moneycontrol. “We have terabytes of commerce data, voice recordings, and consumer interactions that we’ve barely tapped into—and this is our first real step in putting that to work.”

Shunya will be directly integrated into Shiprocket’s seller platform and priced at Rs 499 per month. The company claims early pilots have reduced catalogue and content creation time by up to 40 percent, though it did not disclose customer names or volume benchmarks. A free tier will offer up to 25 queries before sellers need to subscribe.

Shiprocket says the model will help automate tasks like product listings, ad creatives, invoice generation and customer engagement in regional languages. While Shunya.ai is being positioned as a multimodal engine capable of handling voice, image and text inputs, the product remains in very early stages. Key features like voice automation are yet to be rolled out.

“We are proud to power India’s first sovereign AI engine with secure, high-performance GPU infrastructure, fully hosted within the country,” said Seema Ambastha, CEO, L&T Cloudfiniti.

With this, Shiprocket joins a growing list of Indian startups attempting to build vertical AI stacks tailored to local business needs. Unlike open-source or global models, the company says Shunya is purpose-built for India’s regulatory and linguistic landscape, though it did not share technical benchmarks or comparisons with other models.

Ultrasafe, which has previously developed models for healthcare and financial services firms in the US, co-developed Shunya.ai through a joint venture. The development of the model cost around $40 million, according to Ultrasafe CEO Raaid Hossain, who added that it was built “from the ground up with trust, privacy, and scalability at its core, tailored to Indian use-cases.”

The model will also be made available via APIs and low-code tools for external developers and platforms, with integrations planned across India’s digital infrastructure stack including Aadhaar, UPI and ONDC.

Shunya.ai is being positioned as a long-term strategic bet for Shiprocket, which is preparing for a public listing and expanding beyond core logistics. But whether a sovereign AI engine gives it a real edge in a crowded and fast-evolving ecosystem remains to be seen.