A web-based platform can now help power electronics engineers quickly choose semiconductor devices using physics-based analysis. In a discussion with Nidhi Agarwal from EFY, Archit Ashthana from AR Power discusses the development and scaling of his web-based technology platform, focusing on product selection aimed to support engineers…
Q. What does your company do?
A. Our platform is a physics-based analytical engine that helps power electronics engineers select the best semiconductor device in minutes, not months. Engineers input their converter topology and operating conditions, and the platform evaluates thousands of devices from multiple manufacturers to surface the optimal choices. Beyond device selection, we now support integrated thermal modelling, heat sink design, and loss simulation, all within a single platform. No more jumping between datasheets, spreadsheets, and disconnected simulation tools.
Q. What made you build this platform?
A. I spent 8+ years designing power converters across industrial, automotive, and mining applications. I saw firsthand that engineers spend 70% or more of a 2+ year design cycle on component selection decisions, relying on gut feel, incomplete datasheet comparisons, and manual spreadsheet calculations. This approach leads to overdesigned products, higher BOM costs, and missed performance targets. I built AR Power to eliminate that inefficiency entirely by replacing guesswork with rigorous, physics-based analysis at scale.
Q. Who are your target users?
A. Our primary users are power electronics engineers designing any type of power converter across electric vehicles, renewable energy, AI data centre power delivery, industrial motor drives, and consumer power supplies. We are also working on providing free access to Indian university students using their official email IDs to support early learning and adoption.
Q. How does AR Power work, and how are electronics involved in it?
A. AR Power is a web-based platform for power electronics design. The converters designed on our platform are used in electric vehicles, renewable energy systems, AI data centres, and industrial equipment. When an engineer logs in, they select their converter topology, enter their operating conditions, and the platform runs physics-based loss calculations, thermal analysis, and device ranking across our entire database. Results are delivered within minutes. We are also working on direct import from PLECS, LTspice, and Simulink so engineers can skip manual data entry entirely. We use Pareto frontier ranking to identify the best trade-offs among efficiency, cost, and thermal performance. Smart Mode gives advanced users deeper capabilities, including multi-objective optimisation and exportable reports
Q. How do you balance ease of use for engineers with complex physics-based calculations on your platform?
A. The complex physics runs entirely on the backend: loss calculations, thermal impedance modelling, iterative temperature convergence, and Pareto optimisation. On the front end, engineers see clean, actionable results they can use immediately. An entry-level engineer can get meaningful output in minutes, while experienced engineers can use Smart Mode for full analytical control. The platform does the heavy lifting so engineers can focus on design decisions, not spreadsheet work
Q. How accurate is your platform?
A. We have validated our platform against hardware test data and achieved 95 to 96% accuracy in loss and thermal predictions. This level of accuracy is considered highly reliable for engineering design decisions and is comparable to or better than manual calculation methods that take weeks to complete.
Q. How many companies do you currently have in your database?
A. We currently have thousands of devices in our database across multiple semiconductor manufacturers, with more being added continuously through our validation pipeline. Our ingestion system is built to scale. We have a 14-module automated pipeline that extracts device parameters directly from manufacturer datasheets, validates the data, and makes it platform-ready.
Q. How do you decide which semiconductor manufacturers to partner with?
A. Our goal is to be manufacturer-agnostic so engineers get unbiased, best-in-class recommendations. We are open to partnering with manufacturers in two ways. First, API integration, where manufacturers can embed our selection and analysis engine directly into their own design tools, helping their customers make better decisions faster. Second, featured visibility on our platform, where a partner’s devices are highlighted to engineers who are actively evaluating components for real projects. This gives the partnered manufacturers a direct channel to reach design engineers at the moment of decision.
Q. What are the design challenges that you have faced while designing this and how did you resolve those challenges?
A. The biggest technical challenge was turning complex physics-based engineering calculations into a fast, reliable web application. The first version was a working prototype that I tested extensively with engineers to validate the core concept. From there, I brought in web development expertise to build a production-grade frontend while I focused on the analytical engine. On the data side, a major challenge was extracting accurate device parameters from manufacturer datasheets at scale. We solved this by building a 14-module automated ingestion pipeline that handles extraction, validation, and normalisation systematically. On the computation side, getting accurate results requires solving iterative problems like temperature-dependent resistance convergence, which we have optimised to run within seconds rather than the hours it would take manually.
Q. Are you receiving feedback from users? How has the response been?
A. Absolutely. We have over 1000 users on the platform today, all through organic signups with zero paid marketing. Even before launch, I was running structured discovery calls with engineers to understand their workflows and pain points. That process continues. Users are actively requesting new topologies, additional manufacturers, and deeper analysis features. This direct feedback loop drives our entire product roadmap. We are building what engineers actually need, not what we assume they need.
Q. What is your revenue generation model for this product?
A. We will generate revenue mainly from the subscription model. Users will need to pay a certain amount to access the smart mode tool. We will venture into other items in the longer run.
Q. Have you generated any revenue since the launch?
A. We are currently in our growth phase, focused on building a strong user base and refining the product based on real engineer feedback. We have over 1000 users on the platform through organic adoption.
Q. What does your current team look like, especially the design team?
A. The team is lean and focused. I lead product, domain engineering, and go-to-market. We have specialists contributing to thermal design and data science, and developers managing the platform. We are actively looking for a co-founder on the business and marketing side to accelerate growth.
Q. What are the current challenges which you are facing as a startup ?
A. The biggest challenge is changing established workflows. Power electronics engineers have been selecting components the same way for decades, using datasheets, spreadsheets, and experience. Convincing them to adopt a platform-based approach takes time and consistent proof of value. Beyond that, as a bootstrapped startup, we face resource constraints in hiring for development and sales, which we are actively addressing through funding and strategic partnerships.
Q. What is your long-term plan?
A. Our long-term vision is that every power converter designed should be optimal, reliable, and production-ready in its first version. No unnecessary prototyping cycles, no field failures from poor component choices. Today we cover semiconductor selection, loss simulation, thermal modelling, and heat sink design. We are expanding into capacitor selection, magnetics design, and full converter-level optimisation with simulation integration into tools like PLECS, LTspice, and Simulink. That is where we are headed: a world where the first design is the right design.
Q. With full automation, do you think such platforms could replace power electronics engineers in the future?
A. No. AR Power is a tool for engineers, not a replacement. Real-world design will always require engineering judgment: vendor negotiations, supply chain decisions, layout constraints, compliance requirements. What AR Power eliminates is the repetitive, time-consuming analytical work that keeps engineers stuck in spreadsheets instead of doing what they are best at, which is solving hard design problems and innovating. We make engineers superhumans, not redundant.
Q. Are you looking for collaborations or partnerships, including academic tie-ups and channel partners?
A. Yes, we are open to partnerships of all kinds. This includes academic collaborations, where we plan to offer free access to university students worldwide using their official email IDs to support early learning and adoption. We are also open to adding new distributors and channel partners.
Q. What are your plans for future growth, and where do you plan to invest?
A. Our immediate priority is hiring to strengthen the engineering and product team. I am currently handling sales and going to market directly, which gives me a first-hand understanding of what engineers need and how they buy. Once the core platform is fully built out, we will scale sales and marketing aggressively. The goal is simple: build the best product, prove it with customers, then scale globally.
