A robot inspecting an electrical substation cannot afford to miss a fault, just as an autonomous vehicle cannot afford to overlook an unexpected obstacle. Building machines that can operate reliably in such environments requires data, software, testing infrastructure, and hardware that help them interpret the physical world.
Bengaluru-based CaliperAI is developing this infrastructure alongside its own modular robotic platforms.
Founded in 2024 by Mohanvarakrishna Boddeti and Gautham Raju, the startup initially focused on gaps in autonomous-driving software development and validation before expanding into physical AI and unmanned ground vehicles (UGVs) for inspecting heavy industrial assets.
The founders spent six to eight months validating the idea before leaving their jobs to work full-time on the startup in May 2025. CaliperAI currently has a team of eight.
The company’s name reflects its focus on precision.
“We wanted to develop AI which works as precisely as measuring with calipers,” says Co-founder Gautham Raju. “While AI can go wrong anytime, we want to be extremely precise in the AI we develop."
Building software for physical AI
CaliperAI began by addressing a gap in autonomous-driving development. While tools were available for software development and validation, the founders saw a need for an integrated platform that could support the entire development and release cycle.
“There are software development tools and separate tools to validate it, but not an end-to-end intelligent platform that can cater to the complete development and release cycle,” says Co-founder Mohanvarakrishna Boddeti.
The startup has developed three software tools: CaliperCapture, which collects multimodal sensor data; CaliperOwl, which helps developers understand and organise datasets; and CaliperGT, which generates ground-truth labels used to train AI models.
Together, these tools are designed to reduce the manual work involved in preparing data and validating autonomous systems.
“We have built a complete intelligent stack for end-to-end development and validation for physical AI,” Boddeti says. “It need not be just for autonomous robots and can work with any robot that uses sensors.”
The platform uses AI-assisted and agentic workflows to automate parts of data labelling, a process that traditionally requires people to identify objects and situations in sensor recordings.
The aim is to reduce manual intervention while helping developers identify difficult scenarios and assess how their systems respond.
CaliperAI is also using its software expertise to develop a UGV for inspecting heavy assets such as cranes, transformers, transmission lines and electrical substations.
“If you want a human to go to these heavy assets and keep patrolling, it is not safe for them,” says Raju, referring to the risks of working around high-voltage electrical infrastructure. “We do not replace humans, but we help them where it is unsafe for them to work or move around.”
The robot is being developed as a configurable platform rather than one designed for a single task. Its modular architecture could support applications such as warehouse operations, surveillance, patrolling, and electrical-grid monitoring.
Making robotic hardware reusable
CaliperAI wants to bring the reusability of software to robotic hardware. Instead of developing a new machine for every application, the startup aims to adapt existing mechanical, software, and AI components to different environments and requirements.
“We want to modularise physical AI,” Raju says. “With modular software, you develop it once and reuse it based on configurations and architectures. You don't have to keep rewriting the same functions. We want to replicate that for hardware.”
Raju gives the example of a hypothetical solar farm in Kutch, Gujarat, that needs a UGV for monitoring.
Rather than importing a robot and modifying it for the site, CaliperAI wants to develop a suitable configuration and move from testing to deployment within a shorter period.
“In about five months, we should finish a field trial. In six months, we should be in a position to deploy 100 robots in their farm,” Raju says.
The timeline is an illustration of the company's intended development process, rather than an existing deployment.
Funding and plans
CaliperAI is bootstrapped, with the two founders investing about Rs 2 crore of their own money.
The startup is exploring a seed funding round to support hardware development and build physical validation infrastructure, including facilities for testing in off-road environments and areas where GPS signals are unavailable.
“We want to create our own hardware and own the entire chain, and make it modular,” Raju says. “Since it will be modular, we can be extremely cost-effective. Otherwise, people will not be able to afford the robots.”
CaliperAI is part of YourStory's Tech30 cohort—a showcase of India's most promising startups of 2026, unveiled at TechSparks Bengaluru.
Edited by Teja Lele