Anjali Sardana, founder of instant household services startup Pronto, has launched a new venture, Provision Labs, in the United States. The San Francisco-based startup will focus on building infrastructure for physical artificial intelligence (AI), supporting the development of systems that can interact with objects and perform tasks in real-world environments.
Provision Labs aims to work with developers building AI models for robotics and other physical applications. Its focus includes data collection, model evaluation and deployment, areas that are becoming increasingly important as AI moves beyond software-based applications into the physical world.
Provision Labs Focuses on Data for Physical AI
Physical AI refers to AI systems designed to perceive and interact with their surroundings, including robots that can handle objects, navigate environments and carry out practical tasks.
Developing such systems requires large volumes of real-world data to train models and assess their performance in different conditions. Provision Labs aims to support AI developers by helping build the data and evaluation infrastructure needed to develop and deploy these systems.
The company has listed investors including Lachy Groom, General Catalyst, Bain Capital Ventures and Glade Brook Capital. However, its funding amount, valuation and initial commercial customers have not been publicly disclosed.
Sardana Continues to Lead Pronto
Sardana will continue as chief executive of Pronto while leading Provision Labs. Pronto connects households with service professionals for everyday tasks such as cleaning, laundry and meal preparation.
Founded in 2025, Pronto has expanded in India’s instant home services market. In May 2026, the startup extended its Series B funding round to $45 million, with a fresh $20 million investment from Lachy Groom. The round valued the company at $200 million, according to reports.
The launch of Provision Labs marks Sardana’s expansion from consumer services into AI infrastructure, with the new venture targeting developers working on robotics and other systems designed for physical tasks.
Real-World Training Data Creates New Opportunities
The launch comes amid growing demand for real-world training data to improve the ability of AI systems to operate in homes, workplaces and industrial environments.
Pronto has also introduced an opt-in initiative called Pronto Verified, involving wearable cameras to record household service sessions for AI training. The initiative has raised questions about privacy, informed consent and the handling of footage captured inside homes.
Although Provision Labs’ focus overlaps with the broader need for physical AI training data, the companies have not publicly confirmed that the new venture will use recordings collected through Pronto’s programme. Its data sourcing arrangements and commercial partnerships remain undisclosed.
As robotics developers seek reliable ways to train and test AI systems, Provision Labs is entering a growing market for the infrastructure that supports physical AI. Its progress will depend on how it builds its data pipeline, works with developers and converts its technical capabilities into commercial services.
