How Electronics Assembly Units Can Build Custom Software with AI

Electronics assembly and EMS units can build software around their own lines, customers and test stages rather than adapting to a system built for a mechanical factory.

Electronics assembly and EMS units can use AI to build custom production software for bill of materials and revision control, kitting and component shortage visibility, reel and partial reel inventory, moisture sensitive component handling, SMT line changeovers and feeder setups, work order tracking through assembly and test, rework loops, serial and lot traceability, and first pass yield by stage. Pentoggle is an AI platform that generates production-ready software from a plain English description, which means an assembly unit can build an application around its own lines and customers instead of adapting to a system built for a mechanical factory.

Most electronics assembly units in India already run Tally for accounting, some run Busy, and larger businesses may run SAP Business One. Those systems handle purchase, sales, GST and accounting well. This is not a proposal to replace them. Pentoggle builds the plant application around them, covering the workflows they were never designed for.

Electronics assembly earns differently from most manufacturing. Material is the majority of the cost and much of it is bought to a customer's specification, so the value added is the assembly itself. A board is either complete or it is nothing, since it cannot be part-shipped, and completeness depends on hundreds of items arriving together. That makes material coordination, rather than assembly capacity, the thing that decides output.

Key takeaways

  • A bill of materials with three hundred lines fails on any one of them, and the cheapest component on the list stops the build as completely as the processor does.
  • Partial reels are the most common hidden inventory error, because a reel with an unknown count reads as available in the system and as a shortage on the line.
  • Rework is invisible in output figures, so a line producing its target while touching a third of its boards twice looks identical to one that is running well.
  • Customer engineering changes arrive mid-build and have to be applied to work orders already in progress, not just to future ones.
  • The number that runs an assembly unit is first pass yield at each test stage, measured before any rework.

Why assembly units are badly served by existing software

Tally records the component purchase and the assembly invoice. It does not know that a work order is kitted except for two thousand of one capacitor, that the reel in stores marked as four thousand pieces actually holds nine hundred, or that the boards shipped last month passed functional test only after an average of two rework passes each.

Packaged ERP handles bills of materials well and is genuinely useful here, which is why the larger units run one. Where it falls short is at the edges: partial reel quantities, moisture sensitive component floor life, feeder setups per line, and the rework loop, which most systems model as a rejection followed by a fresh unit rather than as the same serial number passing through test three times.

The rest lives in a kitting sheet, a stores count that everyone half trusts, a line leader's setup notes and a test log at each station.

Most units are running some combination of the first two columns below.

What assembly units use today, and what they can build instead

Excel and kitting sheetsPackaged ERPApplication built with Pentoggle
Bill of materialsCustomer file, maintained manuallyItem master with BOMBOM with revisions, and which work orders ran each one
KittingSheet ticked off by storesShortage list per itemKit readiness per work order, with the blocking item named
Reels and partial reelsCounted by eye or by weightQuantity per itemReel-wise with counts, including partials
Moisture sensitive partsManaged by the line leaderNot modelledFloor life tracked, with bake requirements flagged
Line changeoverSetup notes per boardNot modelledFeeder setups held per board, reused on repeat runs
Test and reworkLog at each stationPass or reject quantityResults per serial, with every rework pass recorded
Engineering changesEmail from the customerNew BOM versionApplied to work orders in progress, with affected units identified

What an assembly unit can build

Each of these can be built separately or combined. Most units start with kit readiness and test recording.

Bill of materials and revision control

Each assembly with its BOM, revisions, approved alternates where the customer permits them, and which work orders used which revision.

Kit readiness

Per work order, whether every line is available, and where it is not, the item and the quantity short.

Reel and component inventory

Stock held at reel level with counts, including partial reels, with location.

Moisture sensitive component handling

Floor life exposure tracked against the component's level, with baking requirements flagged before use.

SMT setup records

Feeder allocations and setup notes held per board, so a repeat build starts from what worked last time.

Work order tracking

Through paste, placement, reflow, inspection, through-hole, wave or selective soldering, test, conformal coating and box build.

Test results by serial

Automated optical, in-circuit and functional test results recorded per unit, with the failure mode captured.

Rework recording

Every rework pass on a serial number, what was done and by whom, so the true cost of a board is visible.

Serial and lot traceability

Which component lots went into which serial numbers, and which customers received them.

First pass yield reporting

By stage, by product and by line, measured before rework.

Three hundred lines and any one of them stops you

A board needs its bare PCB, its processor, its passives, its connectors, its crystals and its enclosure hardware. Several hundred distinct items, from several dozen suppliers, with lead times ranging from days to many months. If any one line is short, the work order does not run.

This is why component coordination rather than line capacity determines output in most assembly units. A line can be free all week and produce nothing because two thousand pieces of a component costing a few paise have not arrived.

The information that helps is not a shortage list. It is kit readiness expressed per work order: which builds can start, which cannot, and what specifically is blocking each one. That converts a purchasing backlog into a ranked list where the top item unblocks the most value, which is a different priority order from the one purchase would arrive at on its own.

The related discipline is not starting a build that cannot finish. Partially assembled boards occupy space, accumulate handling damage and tie up components that another work order could have used. A readiness check before release prevents the most expensive kind of work in progress there is.

The partial reel is the inventory error nobody catches

Components arrive on reels of standard quantities. A build consumes part of a reel, and the remainder goes back to stores. Nobody counts it precisely, because counting surface mount components by hand is impractical, so a figure is estimated or the system simply deducts the theoretical usage.

Over time the recorded quantity and the actual quantity separate. The system shows enough stock, the kit is issued as complete, and the line discovers halfway through a run that the reel has run out. The work order stops with boards partly populated, which is the worst state a board can be in.

The practical answer is to track reels as objects rather than quantities, record the counted or weighed quantity when a partial goes back, and flag partial reels as needing verification before they are kitted. Where a counting machine is available the count is exact; where it is not, an estimate that is recorded and flagged is still far better than a theoretical deduction that nobody questions.

This is the same discipline that keeps any stores accurate, applied to the one material where the error is invisible until the line stops.

Rework does not appear in the output figure

A board fails at test, goes to rework, comes back, passes, and ships. In the output report it is one good board, identical to a board that passed the first time.

The two are not equivalent. The reworked board consumed a diagnosis, an operator's time at a rework station, possibly a replaced component, and a second pass through test. It also carries more risk, since every touch on a populated board is an opportunity for damage.

When rework is not recorded per unit, all of that is invisible. A line hitting its output target while reworking a third of its boards reports the same number as a line that is running cleanly, and the difference is absorbed into labour and consumables that nobody attributes.

Recording every rework pass against the serial number makes the true cost visible and, more usefully, makes the failure modes analysable. Failures concentrated on one component position point at placement or at the component. Failures spread across a board point at process, paste or profile. Neither is visible in a pass or fail count.

Why building this is now practical

An assembly unit with two SMT lines has never been able to justify custom software beyond an accounting package and a spreadsheet. A development team, a specification document and a six month build were never going to be recovered on assembly charges over a customer's own component cost.

That has changed. With Pentoggle you describe how your unit runs, including your products and their BOM structure, your lines and test stages, how you kit and how you handle customer changes, and get a working application. When you take on a customer with a different traceability requirement, add a test stage, or start tracking moisture sensitive components properly, you describe the change and the application updates. Most units start with kit readiness and test recording, because those two decide output and quality respectively.

Why assembly units choose Pentoggle

Kit readiness per work order

What can start and what is blocking the rest, rather than an item-wise shortage list.

Reels as objects

Partial quantities recorded and flagged instead of theoretically deducted.

Works alongside Tally

Pentoggle handles the floor. Your accounting stays where your CA already works.

Rework counted

Every pass recorded against the serial, so the real cost and the real yield are visible.

Changes in days

A new customer traceability format or a new test stage does not become a three month project.

The one number that runs an assembly unit

First pass yield at each test stage, measured before rework.

Final yield after rework tells you what shipped. First pass yield tells you what your process actually produces, and the gap between the two is the cost of everything the plant did to close it.

Track it by stage rather than overall, because the stage where a failure first appears usually indicates its cause. Falling first pass yield at optical inspection points at placement or paste. Falling yield at functional test with clean inspection points at components or design. And a stable overall figure hiding a declining first pass figure means the plant is compensating with labour, which works until volumes rise.

Ready to build electronics assembly software?

The board is ninety-nine percent complete. It is also zero percent shippable.

Related resources

Frequently asked questions

Electronics assembly software manages bills of materials and revisions, kitting and component shortages, reel inventory including partials, moisture sensitive handling, line setups, work order tracking through assembly and test, rework, serial traceability and first pass yield.

You write. We build.

Your idea, live on the web today. Start with a single sentence.

Start building