Multi-Phase BuildProject duration
QA, store, and production intelligence across the shop floorScope
Metal Parts ManufacturingIndustryOverview
Mourya Global is a metal parts manufacturer that runs everything in house, from the foundry through production, quality assurance, and the store. The team knows its work deeply, but the tools they run on are dated. Most data still lives on paper, decisions run on memory and experience, and management has no clear real-time view of the shop floor. What they needed was not another ERP, but a layer of intelligence sitting on top of the operations they already run. That is what we built for them. The solution works as an intelligence layer above the factory, made up of a few department-focused AI agents feeding one shared dashboard. It leaves the way the factory already works untouched and gives management the live view it never had.
The Challenge
On the floor, the gaps were everywhere. QA depended on manual measurement sheets, with no reliable way to track process capability or rejection trends. The store ran on institutional memory, with manual counting, hand-entered invoices, and lookalike materials that often got mixed up. Production tracking sat mostly in supervisors' heads. Knowledge lived in people, not in the system, so the moment someone was busy or absent, the data went with them. Mourya needed a solution that could: • Give management a live view of the shop floor instead of a paper trail that arrives late • Catch QA problems as measurable trends, not gut feel, and flag defects the eye misses • Bring the store under control, so stock, invoices, scrap and returns are recorded as they happen • Track every production order live and raise a flag the moment something drifts • Sit on top of the existing operation without forcing the team to change how they work The real challenge was capturing knowledge that had always lived in people's heads, and doing it without slowing down a floor that already knew its job.
What We Built
The QA agent. It checks measurements against the drawing, calculates process capability automatically, and logs every rejection so trends become visible instead of anecdotal. A camera-based system sits alongside it, spotting visual defects like cracks and porosity on every part, without tiring the way a human inspector does. The store agent. Each item is QR-tagged and verified against the order, so lookalike materials stop getting mixed up. It reads invoices straight into inventory, and records scrap and returns through simple conversation instead of hand entry. The production agent. It tracks every order live and raises an alert the moment a machine goes idle or a parameter drifts, so a problem surfaces while it can still be fixed. One live dashboard. Everything the agents capture feeds a single real-time view of the floor, backed by an advisory layer that answers process questions in plain language, drawing on Mourya's own history along with wider technical knowledge.
Why It Wins
• An intelligence layer, not a rip-and-replace. It sits on top of the operations Mourya already runs, rather than forcing a new ERP onto the team. • Built around this factory. It is shaped to Mourya's actual furnaces, alloys, and rejection patterns, not a generic template. • It gets better over time. Because it learns from the plant's own history, it captures lost data early and steadily reduces rejections as it goes. • Eyes that do not tire. Camera-based defect detection checks every part, catching what a tired inspector at the end of a shift would miss. • Knowledge that stays in the system. What used to live only in supervisors' heads is now captured as it happens.
Outcome
We built an intelligence layer that gives Mourya Global a live view of its own shop floor for the first time, with QA, store and production each under a dedicated agent and everything visible on one dashboard. Because it learns from the plant's own furnaces, alloys and rejection patterns, it captures data that used to slip away and reduces rejections as it runs.
More projects
Let’s build something that matters
Have an idea, a problem to solve, or a product you want to improve?
