Custom software in Ipoh: 3 build approaches a Perak factory can take in 2026

A 40-person Ipoh metal-stamping shop, an 80-person Sitiawan ceramic-tile plant, a 25-person Taiping rubber-glove line, a 60-person Gopeng limestone-quarry operator — the calls from Perak factories all sound the same. The owner is staring at a paper job card or a half-broken Excel sheet, knows off-the-shelf software doesn't fit a Malaysian factory floor, and wants a straight answer on custom software in Ipoh before they sign anything. There are 3 honest build approaches in 2026, and the right one depends on which problem you're actually trying to fix.

What this is and why it matters

Perak's industrial base is older and more spread out than Selangor's. Ipoh's old tin-mining belt has been recycled into ceramics, metal stamping, limestone quarrying, and a long tail of 30-to-80 person factories that feed the Klang Valley assemblers. The factory owner asking about custom software in 2026 is not asking about a "transformation" — they have one specific thing that's costing them money, and they want it fixed. A mis-scoped build in this segment is worse than no build at all: a 40-person stamping shop that signs a RM 280k "platform" contract will end up with a half-finished dashboard, a worn-out accounts team, and a 12-month delay. The AI agency Malaysia picture sits on the head page; what follows is the Perak-specific build-path read for a real factory, not a procurement deck.

The 3 build approaches a Perak factory can take in 2026

1. Tier 1 — Pre-built AI, lightly customised (RM 500 – RM 5,000 setup + RM 50 – RM 500 a month). The cheapest legitimate answer, and the right call for about a third of Perak factories that call us. Take a SaaS or open-source tool that already does 80% of the job — a Make.com + Claude flow that summarises daily production reports, a Roboflow vision model on a small SKU set, a HubSpot AI workflow that drafts customer follow-up emails in Bahasa Malaysia — wire it up to the factory's existing data, hand it over in 1-2 weeks. The honest read: if your problem is "we need better visibility into one specific thing," this is the move. The Tier 1 Perak custom-software services page goes deeper on which factories should stop here and not graduate. Examples we ship: a 30-person Ipoh stamping shop automating its daily rejection log into a WhatsApp digest; a 50-person Sitiawan tile plant wiring up an off-the-shelf OCR reader to log pallet labels at the dispatch dock.

2. Tier 2 — Single-workflow custom build (RM 25,000 – RM 80,000). Where the bulk of real Perak factory builds should land, and what we'd recommend to about half the owners who call us. One workflow, one user group, one measurable payback. The classic examples: a custom work-order tracking system for a 60-person Gopeng quarry that links the loader scale to the lorry queue; a 40-person Ipoh stamping shop's in-process inspection log tied to its press uptime; a custom quotation-to-cash pipeline for an 80-person Sitiawan ceramic-tile supplier who still quotes in WhatsApp and Excel. Build takes 4-10 weeks, payback is usually 3-9 months, and the owner owns the code at handover. This is the Perak AI agency hub sweet spot — the build cost lands at "one hire for a year" rather than "one project." The hidden cost most Perak factories miss at Tier 2 is the data-cleanup sprint: 2-4 weeks of SKU master cleanup, operator log normalisation, and ERP export mapping. Budget RM 8k-20k for it. If a vendor quotes you without asking about data quality, they're a salesperson.

3. Tier 3 — Multi-workflow platform (RM 80,000 – RM 300,000+). The right call for a Perak factory that has outgrown Tier 2, has 3-5 workflows that genuinely need stitching, and has the internal discipline to drive a 3-6 month build. The honest version: about a third of Perak factory owners who ask for a Tier 3 build should be talked back down to Tier 2. The platform pitch is rarely the right first move — the factory that needs order-to-dispatch plus inventory plus invoicing usually needs one of those three done well first, then the other two layered in. Examples that actually fit Tier 3: a 120-person Ipoh metal-stamping group with 3 plants, 2 ERP exports, and a customer-portal ask; a 200-person Sitiawan tile exporter with a 6-country dealer network that needs an order-to-cash system. Build takes 3-6 months, requires a dedicated factory-side project owner (not just the owner driving it on weekends), and the vendor shortlist in Perak for genuine Tier 3 work is small — most Perak agencies will either sub-contract or quietly downscope you to a Tier 2.

Where it goes wrong

Three failure modes we see most in Perak factory builds. First, the offshore-cheap path — a few Penang or KL-based agencies quote RM 15k-25k for a Tier 2 build that should cost RM 50k-70k. Build lands 4 months late, code is unmaintainable, the factory ends up paying us RM 60k to rebuild. Second, the "transformation" pitch at Tier 3 prices for a Tier 1 delivery — a 40-person Ipoh shop gets sold a RM 200k platform, ends up with a single half-finished dashboard, and the vendor's gone in 6 months. Third, the data-is-fine assumption — the vendor doesn't audit the operator logs, the SKU master, the ERP export shape before quoting, and discovers 4 weeks into the build that the data is in 4 inconsistent formats across 3 shifts. The fix: spec-first discovery with a fixed price, you own the code, one workflow at a time, and a 6-12 month retainer (RM 1,500 – RM 8,000 a month) for retraining and the inevitable "we need it to also do X" after go-live. Perak factories that skip the retainer usually call us back at month 9 asking us to fix what a 1-day-a-month retainer would have prevented.

What to do next

If you're a Perak factory owner looking at a real build, the first move is a 2-4 week discovery engagement (RM 5,000 – RM 15,000, credited against the build). The deliverable is a build spec, not a sales deck — and if the answer is "your problem isn't custom, fix the SaaS stack first" or "wait 6 months, your data isn't ready," we'll say that. About a third of the time the answer is "no, not yet." The Perak custom-software services overview walks through the 3 build approaches with Perak-specific factory examples, and the Perak AI agency hub has 4 representative builds from Ipoh, Taiping, Sitiawan, and Gopeng to anchor against.