Pronto’s Anjali Sardana Buys Basilisk Industries to Bolster Physical AI Hardware

Anjali Sardana, founder of the autonomous vehicle startup Pronto, is expanding her latest venture, Provision Labs, into hardware with the acquisition of Mumbai-based Basilisk Industries. The deal, announced via Sardana’s LinkedIn post and first reported by *inc42*, brings on Basilisk’s founder Shourya Watwani and his team, which has developed perception technologies—including a 3D hand-tracking model and a tactile glove—to help machines interpret and interact with the physical world.

Provision, launched in San Francisco earlier this year, has been building a data platform for training physical AI systems, but its ambitions now extend into the hardware layer. Watwani will lead Provision’s hardware efforts, focusing on in-house post-processing models and integrating Basilisk’s work into Provision’s perception stack. The move indicates a direct response to the challenges of deploying AI in real-world robotics, where tactile feedback and environmental understanding remain key hurdles.

When we covered Provision’s launch on 26 October, Sardana framed the startup as a data infrastructure play for physical AI, collecting training data from service professionals—a model that drew scrutiny over privacy concerns during Pronto’s earlier household recording pilot. Now, the hardware acquisition signals a shift toward owning more of the full stack, from data collection to sensor design. This approach contrasts with competitors like Snabbit, which has partnered with data providers for simulated training, or Urban Company, which has avoided in-home recording entirely.

The deal highlights the growing divide between software-first AI companies and those racing to control the physical layer. Basilisk’s team, founded in 2023, specialized in perception tech—a critical but often overlooked component for robots and wearables. Their 3D hand-tracking and tactile glove prototypes suggest Provision’s focus on dexterous manipulation, a priority for logistics, healthcare, and service robotics. Inc42 reported that Provision is already hiring for roles in computer vision, tactile data, and environment reconstruction across San Francisco, Toronto, and Bengaluru, signaling a global push to scale its hardware research.

The acquisition also raises questions about how Provision will reconcile its data-driven approach with hardware development. Sardana’s earlier work at Pronto highlighted the ethical risks of in-home data collection, and Provision’s platform still relies on crowdsourced input from professionals. If the startup’s physical AI systems require custom sensors or gloves, it may face new privacy and consent challenges—especially if those devices collect data from users’ interactions. Competitors like Snabbit have avoided this path by focusing on simulated environments, while Urban Company’s avoidance of in-home recording suggests a more cautious approach to user-facing hardware.

For founders and investors watching the physical AI space, the Basilisk deal serves as a test case for whether data alone can power robust robotic systems—or if hardware ownership is necessary to compete. Provision’s hiring spree and the integration of Basilisk’s team suggest it’s betting on the latter. The next step will be whether the startup can turn its tactile and perception tech into commercial products before other players in the field establish their own hardware advantages. If successful, Provision could redefine how physical AI is trained—not just with data, but with sensors that directly influence the models themselves.

Sources: inc42.com