In December 2024, I left a 14-year career as Technical Manager at a packaging manufacturing company to start Peribott Dynamic LLP. I was building an autonomous mobile robot company, in Hyderabad, with no outside funding, no co-founder, and a very specific conviction: that most AMRs fail in production because they are built by people who have never operated production machinery.
Twelve months later, devibot exists. It navigates. It maps. It boots in under 90 seconds without anyone touching a terminal. The dashboard runs on a touchscreen and does not require the operator to know what ROS2 is.
Here is the honest account of the year. Not the LinkedIn version.
What nobody tells you about robotics hardware in India
The import situation for robotics components in India is genuinely difficult. High-quality motor controllers, LiDAR sensors, and industrial-grade hardware are either import-only, expensive, or both. Our LD19 LiDAR, Syntron motor drives, and Pylontech BMS all came through import channels. Lead times are unpredictable. Customs clearance adds uncertainty you cannot plan around.
This matters because hardware timelines drive everything else. You cannot test navigation software without a working drivetrain. You cannot integrate the BMS without the BMS in hand. Software development is fast; hardware procurement is slow. Every AMR project’s critical path runs through customs.
The practical adaptation: order hardware two to three weeks earlier than you think you need to. Keep backup units for critical components. For devibot, we maintain spare motor controllers and a spare LiDAR unit on-site. A single failed component shutting down development for three weeks is a cost that justifies the inventory.
The advantage I did not expect: 14 years of production experience
I spent 14 years in packaging manufacturing. Product development, mold design, machine maintenance, equipment upgrades — the unglamorous operational side of manufacturing. I did not expect this to be an advantage in robotics. It turned out to be the most important thing I brought to devibot.
Because I know how machines fail in production. Not in theory — in practice. The NaN that crashed our dashboard, the ISR mutex that caused intermittent HardFaults, the boot sequence that left robots in undefined states — I recognised these failure patterns because I had seen their analogues in packaging machinery. Machines that work perfectly in acceptance testing and fail on the shop floor because nobody designed for the edge cases that only appear after 500 hours of continuous operation.
This informed every design decision in devibot. The auto safe-mode recovery in the boot system, the NaN sanitisation pipeline, the graceful shutdown sequence, the hardware E-stop integration — none of these are obvious features for a first robotics product. They are obvious features if you have spent 14 years watching production systems fail.
The market is real, and it is underserved
International AMR vendors — MiR, Omron, Geek+ — price their robots in the ₹25–50 lakh range. For large enterprises with significant automation budgets, this is viable. For the SME manufacturers in Hyderabad’s industrial clusters — Patancheru, Pashamylaram, Nacharam — it is completely inaccessible.
These manufacturers have the same logistics problems that AMRs solve. Repetitive material transport between fixed points. Predictable floor layouts. High labour costs for low-skill repetitive movement. The economics of automation make sense; the price point of imported AMRs does not.
Building in India, at Indian cost structures, with local support, positions Peribott to serve this segment. A robot at ₹8–12 lakh with a payback period under 24 months is a different conversation than a ₹40 lakh import that requires specialised integration services.
What I would do differently
Get a reference customer earlier. The first version of devibot worked before we had a paying customer committed. A reference customer creates urgency, provides real-world feedback, and produces the case study that becomes your most effective sales tool. I would have secured a pilot commitment before completing the full product.
Separate firmware development from software development more strictly. STM32 firmware bugs and ROS2 software bugs interact in ways that make debugging extremely difficult. A sensor returning garbage data might manifest as a navigation failure that looks like a software bug. Cleaner interfaces between layers, with explicit fault reporting at each boundary, would have saved significant debugging time.
Document earlier. The AMR Boot System, the dashboard, the fleet platform — all of these evolved through dozens of iterations. Early documentation forces you to articulate design decisions while you still remember why you made them. This article is partly an attempt to catch up on that debt.
What comes next
Peribott is in active conversations with potential first customers in Hyderabad’s manufacturing sector. The technical foundation is solid — devibot navigates reliably, the platform is production-grade, and the fleet management platform is ready for multi-robot deployments.
The next challenge is commercial: converting technical credibility into signed contracts. That is a different kind of problem from building the robot, and a different set of skills. The engineering part I know how to do. The selling part I am learning.
If you are a manufacturer in Hyderabad looking at automation, or an investor interested in Indian robotics, I am genuinely interested in talking. Not to pitch — to understand the problem from your side.
Amit Jagnani is the Founder & CTO of Peribott Dynamic LLP, Hyderabad. Reach out directly — he reads every email.