Monday plan, Thursday reality
A machine goes down, a delivery slips, and nobody replans because replanning by hand takes longer than the day has left.
Manufacturing
Manufacturing software development, written from the shop floor inwards
Production planning that survives a breakdown. Machine data read from the PLC instead of a logbook. Traceability that answers a customer complaint in minutes, and costing that reflects what the job actually consumed.
5 solution blueprints
15 systems we integrate
10–24 weeks, typical build
The starting point
What is usually broken before we arrive
Production is reported from memory
Output, rejections and downtime get typed up at the end of the shift. Every number built on that inherits the guesswork.
Traceability lives in a filing cabinet
A customer asks which lot, machine and operator made their part. Pulling paper inspection records takes days, and the account notices.
In plain terms
Manufacturing software development is building what a factory plans and runs production on. Production planning and scheduling, shop-floor execution, machine and downtime monitoring, quality control and traceability, stores and BOM-driven procurement, and the job costing that shows what a batch really cost to make.
Ask a plant manager what their software problem is and the answer is about the plan. It is made on Monday from the order book, and by Thursday it bears no relation to what is being made. Software only helps if it can absorb reality fast.
So we start with data capture, not scheduling logic. If job start, job end, quantity, rejections and downtime are entered on a shared PC at shift end, the plan always works from yesterday. Capturing them at the machine is what makes everything downstream possible.
Traceability is the third theme, and customers drive it more than factories do. An automotive or pharmaceutical buyer wants to know which raw material lot, machine, operator and inspection record produced the part in their hand. Answer within the hour and you keep the account.
What we build
Manufacturing systems, and who builds each one
Every card links to the practice that does the work. Follow one if you want the detail.
Production planning and MES
Order-to-schedule planning with machine and manpower constraints, shop-floor job cards, live production reporting, and replanning that takes minutes.
ExploreMachine monitoring and OEE
Data straight from PLCs over OPC UA or Modbus. Availability, performance and quality computed continuously, with downtime reasons captured at the machine.
ExploreManufacturing ERP: BOM, MRP and stores
Multi-level bills of material, MRP against the production plan, stores issue and return, subcontracting, and job costing from actual consumption.
ExploreQuality management and traceability
Incoming, in-process and final inspection plans, non-conformance and CAPA workflow, and lot-to-part genealogy searchable from the part number backwards.
ExplorePlant and multi-plant dashboards
OEE by machine and shift, rejection Pareto by cause, on-time delivery, inventory ageing and per-unit cost trends. Fed from the shop floor, not compiled monthly.
ExploreWorkforce, shift and contract labour
Shift rosters, biometric attendance, contractor gate passes, skill matrices against machines, and payroll that handles overtime and statutory deductions.
ExploreDealer, service and spares portals
Order placement and warranty registration for the dealer network, service ticket tracking, spares availability. The channel side most systems leave out.
Explore Solution blueprints
Open one and see exactly what gets built
The situation it answers, the architecture, every module, the systems it plugs into and how long it takes. These describe our work, not one client's project. Named clients with measured results are in the case studies.
01 Production planning and shop-floor execution 16–22 weeks
The situation
An engineering job shop or batch manufacturer with fifteen to sixty machines, planning from the order book into a spreadsheet. Job cards are paper. Production is reported at shift end. The plan gets revised only when somebody notices it has broken. Delivery dates are promised without knowing true capacity.
How it is built
A finite-capacity scheduling service holds machine, tooling and manpower constraints. Shop-floor terminals capture job events by barcode scan. Node.js and PostgreSQL, with a replan operation fast enough to run whenever reality changes.
Rules it has to follow
Subcontracting moves on job-work challans with material reconciled on return. That is exactly what a GST assessment asks for. Delivery challans and e-way bills are raised at dispatch instead of reconstructed at month end.
Modules 12
- Order book with promised dates and priorities
- Routing and operation master per part
- Finite-capacity scheduling with machine and tooling constraints
- Job card generation and barcode-scanned shop-floor terminals
- Real-time production, rejection and rework reporting
- Downtime capture with reason codes at the machine
- Operator and machine assignment with skill validation
- Work-in-progress tracking between operations
- Subcontracting dispatch and receipt against challans
- Replanning on breakdown or material delay
- On-time delivery and schedule adherence reporting
- Actual versus standard time and cost by job
Plugs into
- Tally ERP
- GS1 barcode standards
- Biometric attendance devices
- GST e-way bill (NIC)
02 Machine monitoring and OEE 10–16 weeks
The situation
A plant where utilisation is estimated from output and downtime is written into a logbook from memory. Management wants OEE, but operators are the ones asked to record losses that make them look bad. The numbers stay optimistic in a consistent direction.
How it is built
Edge gateways read PLCs and controllers over OPC UA and Modbus TCP, publishing to an MQTT broker and a time-series store. Machines without a controller interface get a retrofit sensor plus a terminal for reason-code entry, because a sensor detects a stop but cannot explain it.
Rules it has to follow
Machine data is collected read-only across a segmented network, so the control layer is never exposed to the enterprise network. That is where any operational technology security review starts.
Modules 12
- Edge gateway with OPC UA and Modbus TCP polling
- Retrofit sensor option for legacy machines
- Time-series storage of machine state and cycle counts
- Availability, performance and quality computation
- Downtime reason capture at the machine with a short code list
- Micro-stoppage detection and aggregation
- Shift, machine and product-wise OEE views
- Loss Pareto by reason and by machine
- Andon display on the shop floor
- Threshold alerting to supervisors
- Preventive maintenance triggered by running hours
- Energy consumption per machine and per unit produced
Plugs into
- OPC UA
- Modbus TCP
- MQTT broker
- Power BI Service
- Existing SCADA historians
03 Manufacturing ERP: BOM, MRP and costing 18–24 weeks
The situation
A manufacturer running purchase, stores and accounts on Tally, with production managed separately. Material shortages are found when a job reaches the machine. Purchasing happens reactively at whatever price is available. The cost of a finished part is a standard figure nobody has revisited in two years.
How it is built
A manufacturing core with multi-level bills of material and routings runs MRP against the production plan and on-hand stock. It posts summarised financial entries to Tally instead of duplicating the ledger. Costing accumulates material, labour and machine time per job as work is reported.
Rules it has to follow
Job-work material sent to subcontractors is tracked against challans and reconciled on return inside the period GST allows. Input credit is matched to GRN and quality clearance, so rejected material does not silently carry credit.
Modules 12
- Multi-level BOM with revisions and effectivity dates
- Item master with alternates and approved suppliers
- MRP run against plan, on-hand and open orders
- Purchase requisition, order, GRN and quality clearance
- Stores issue against job with shortage visibility
- Batch and serial tracking through stores
- Subcontract material issue and reconciliation
- Job costing: material, labour, machine and overhead
- Scrap, rework and yield accounting
- Physical stock count with variance approval
- Supplier rating on quality and on-time delivery
- Cost variance reporting against standard
Plugs into
- Tally ERP
- SAP Business One
- GST e-invoicing (NIC IRP)
- GST e-way bill (NIC)
04 Quality management and lot traceability 12–18 weeks
The situation
A supplier to automotive or pharmaceutical customers who must answer a field complaint with the full history of one part. Inspection records sit on paper in the QA room. Raw material lot numbers are recorded at goods receipt and never carried forward. A complaint triggers two days of file retrieval.
How it is built
A genealogy model links raw material lot, machine, operator, shift and inspection record to every produced serial or batch. Capture happens through the same shop-floor terminals used for production, so quality data is a by-product of work instead of a second data-entry job.
Rules it has to follow
Record structure follows what ISO 9001 and IATF 16949 audits examine. Calibration history, non-conformance disposition, CAPA closure evidence, and traceability from finished part to raw material lot, each with the identity of whoever recorded it.
Modules 12
- Inspection plans for incoming, in-process and final stages
- Sampling rules and measurement capture with tolerances
- Gauge and instrument calibration register
- Lot-to-part genealogy across operations
- Non-conformance reporting and disposition
- CAPA workflow with owner and closure evidence
- Customer complaint register linked to genealogy
- 8D report generation from the complaint record
- Statistical process control charts by characteristic
- Supplier incoming quality history and rating
- Traceability search from part number backwards
- Audit evidence pack export
Plugs into
- Digital gauges and measurement instruments
- GS1 barcode standards
- Power BI Service
- Customer supplier portals
05 Dealer, warranty and service network portal 14–18 weeks
The situation
An equipment or consumer-durables manufacturer selling through dealers and servicing through franchised centres. Warranty claims arrive as scanned forms. Spares availability is a phone call. Nobody has a reliable view of which components fail most often in the field.
How it is built
A dealer and service portal in React, with an Android technician app. Behind it, an order and warranty service links every serial number to its dealer, its warranty terms and its full service history from the day it left the plant.
Rules it has to follow
Warranty terms are held against the serial number as sold, so a claim is assessed against the contract that applied on that date, not the current policy. Field failure data links back to production lot, which makes a recall decision defensible.
Modules 12
- Dealer onboarding with territory and credit terms
- Order placement with scheme and slab pricing
- Serial number allocation and dispatch tracking
- Warranty registration at point of sale
- Service ticket creation, assignment and closure
- Technician app with job checklist and photographs
- Spares catalogue, availability and dealer ordering
- Warranty claim submission, review and settlement
- Field failure analysis by component and batch
- Annual maintenance contract tracking and renewal
- Dealer performance and service SLA reporting
- Feedback capture linked to the service ticket
Plugs into
- Tally ERP
- GST e-way bill (NIC)
- Razorpay payment links
- WhatsApp Business API
- GS1 barcode standards
Regulation
What manufacturing software has to comply with in India
Get these wrong and the build is not late, it is unusable. They are design inputs on day one.
ISO 9001 and IATF 16949 records
Audits examine evidence, not intent. Calibration history for every measuring instrument, non-conformance disposition with authority, CAPA closure with proof, and traceability from a finished part back to its raw material lot. Systems that generate these from daily work pass more easily.
GST job work and e-way bill
Material sent to a subcontractor moves on a job-work challan and must be reconciled on return inside the period GST allows, or it becomes a deemed supply. Consignments over the value threshold need an e-way bill carrying transporter and vehicle details captured at dispatch.
OPC UA and Modbus machine interfaces
Most shop-floor data lives in PLCs speaking OPC UA or Modbus TCP, and older machines may offer neither. A monitoring build has to handle both, plus a retrofit path for machines with no digital interface at all. Otherwise it covers only the newest third of the plant.
Operational technology network segmentation
Machine controllers were never designed to sit on a network reachable from the office. Data collection runs read-only through a segmented network with a one-way path to the enterprise side. Any operational technology security review checks this first.
Factories Act shift and overtime records
Shift rosters, overtime hours, weekly rest and contract labour registers are statutory records open to inspection. Payroll that computes overtime and deductions from captured attendance, instead of a monthly summary, keeps those registers defensible.
Why Naryon Tech
How we approach manufacturing projects
- Data captured at the machineBarcode scans and direct PLC reads at the point of work. Production and downtime typed up at shift end are remembered, not recorded, and every number built on them inherits the error.
- Replanning that takes minutesA schedule is only useful if it absorbs a breakdown or a late delivery the same day. The replan runs on demand, so the plan on the floor stays connected to what is happening.
- Traceability from the part backwardsLot, machine, operator, shift and inspection record link to every part produced. A customer complaint resolves in minutes and an audit request resolves without pulling files.
- The control network stays isolatedMachine data is read through a segmented, read-only path. Collecting production numbers should never open a route from the office network to a PLC.
Tooling
What we build manufacturing systems with
Application
- React
- Node.js
- TypeScript
- React Native
Shop floor
- OPC UA
- Modbus TCP
- MQTT
- Barcode terminals
- Edge gateways
Data
- PostgreSQL
- TimescaleDB
- Power BI
- Azure Data Factory
Business systems
- Tally ERP
- SAP Business One
- GST NIC APIs
Working from Chennai and Trichy, with clients in
- Chennai
- Tiruchirappalli (Trichy)
- Coimbatore
- Madurai
- Salem
- Erode
- Tirunelveli
- Vellore
- Tiruppur
- Puducherry
Straight answers
Manufacturing: what people ask us first
Can you collect data from older machines with no interface?
Yes, through a retrofit path. Machines with a PLC are read over OPC UA or Modbus TCP directly. Machines without one get a sensor on the cycle, usually proximity or current, enough for count and running state. A terminal alongside captures the downtime reason, because a sensor detects a stop but cannot explain it.
Do we have to replace Tally to get a manufacturing system?
No, and usually you should not. Tally stays the financial book of record while production, stores and costing run in the manufacturing system, posting summarised entries across on a schedule with a reconciliation report. Replacing accounting during a shop-floor rollout means two disruptions at once.
How accurate is OEE from automated collection?
Availability and performance become accurate almost immediately, because they come from machine state and cycle counts instead of recollection. Quality depends on rejections being reported honestly at the machine. That is a floor-culture question too. So we keep reason codes short, entry fast, and show operators their own shift data.
Will this work across multiple plants?
It is built that way. Item, BOM and routing masters are shared with plant-level overrides where processes genuinely differ, while production, stock and costing are scoped per plant. Group reporting consolidates without forcing every plant onto the same machine list, which is rarely realistic.
How long before the shop floor is using it?
Machine monitoring shows something useful in four to six weeks, because it needs gateways and dashboards instead of process change. Planning and execution take sixteen to twenty-two weeks, and the constraint is usually routing and standard time data that many factories have never written down. We start collecting that in week one.
What happens to the terminals if the server goes down?
Terminals queue job events locally and sync when the connection returns, so production reporting does not stop with the network. Machine gateways buffer readings at the edge for the same reason. The plan and reporting catch up afterwards, which is acceptable in a way that a stopped line is not.
Close to this
Related industries
Building something for manufacturing?
Tell us the workflow that keeps breaking. We will say honestly whether software fixes it, and what it would take.