For reliable operation in complex environments, such as electrical substations, where missing defects is unacceptable, machines capable of interpreting the physical world are necessary. This requires data, software, testing infrastructure, and hardware.
CaliperAI, a Bangalore-based company, is developing this infrastructure in parallel with its own modular robotic platforms.
The startup was founded in 2024 by Mohanvarakrishna Boddeti and Gautam Raju. Initially, the company focused on addressing gaps in autonomous driving software development and validation before expanding its scope to physical AI and Unmanned Ground Vehicles (UGVs) for inspecting heavy industrial sites.
The founders spent six to eight months validating the idea, after which they left their previous jobs in May 2025 to dedicate themselves fully to the startup. The CaliperAI team currently consists of eight people.
The company name reflects its emphasis on precision. Co-founder Gautam Raju notes: 'We wanted to develop AI that works as accurately as a caliper measurement. Although AI can fail at any moment, we want to achieve exceptional accuracy in the AI we develop.'
CaliperAI began by solving a problem in autonomous driving development. Despite the availability of tools for software development and validation, the founders saw a need for an integrated platform that could support the entire product development and release cycle.
Mohanvarakrishna Boddeti, co-founder, emphasizes: 'There are tools for software development and separate tools for its validation, but there is no comprehensive intelligent platform capable of satisfying the full development and release cycle.'
The startup has created three software tools: CaliperCapture for collecting multimodal sensor data; CaliperOwl, which helps developers understand and organize datasets; and CaliperGT, which generates ground truth labels used to train AI models. These tools together aim to reduce manual labor associated with data preparation and autonomous system verification.
Boddeti states: 'We have built a complete intelligent stack for end-to-end development and validation of physical AI. It does not necessarily have to be used only for autonomous robots and can work with any robot that uses sensors.'
The platform utilizes agentic and AI-assisted workflows to automate part of the data labeling process—a process that traditionally requires humans to identify objects and situations in sensor recordings. The goal is to reduce manual intervention while assisting developers in identifying complex scenarios and assessing their systems' reactions.
CaliperAI also applies its software expertise to create a UGV designed for inspecting heavy assets such as transformers, power lines, cranes, and electrical substations. Raju discusses the risks of working around high-voltage infrastructure: 'If you want a human to patrol these heavy assets, it is unsafe for them. We are not replacing people, but helping them where it is unsafe for them to work or move.'
The robot is being developed as a customizable platform, not a device intended for one specific task. Its modular architecture can support applications such as warehouse operations, surveillance, patrolling, and power grid monitoring.
CaliperAI aims to bring software reusability into robotic hardware. Instead of building a new machine for every task, the startup intends to adapt existing mechanical, software, and AI components to various conditions and requirements.
Raju explains: 'We want to modularize physical AI. With modular software, you develop it once and reuse it based on configurations and architectures. You don't have to constantly rewrite the same functions. We want to replicate this for hardware.'
Raju cites a hypothetical example of a solar farm in Kutch, Gujarat, requiring a UGV for monitoring. Instead of importing a robot and modifying it for that site, CaliperAI wants to develop a suitable configuration and move from testing to deployment in a shorter timeframe. He predicts: 'In about five months, we should complete field testing. In six months, we should be ready to deploy 100 robots on their farm.'
This timeline illustrates the company's projected development process, not current deployment. CaliperAI is self-funded, with the two founders investing about 2 crore rupees. The startup is considering seeking seed funding to support hardware development and the creation of physical validation infrastructure, including off-road and GPS-denied testing areas.
Raju concludes: 'We want to build our own hardware and control the entire chain, making it modular. Because it will be modular, we can be very cost-effective. Otherwise, people won't be able to afford these robots.'
CaliperAI is included in the Tech30 cohort from YourStory—a showcase of India's most promising startups in 2026, presented at the TechSparks event in Bangalore.