Malaysia has been pushing construction to digitise for a decade. The government announced that public projects worth RM100 million and above would be required to use building information modelling from 2019, and the Construction Industry Development Board later recommended extending the requirement to certain private projects. A 2019 report put BIM adoption in Malaysia at 17%, against 65% in Singapore. That figure is dated now, but the gap it describes has not closed quickly.
The pattern is familiar. A mandate at the top of the market, slow uptake below it. Most contractors and subcontractors are small firms. A 3D model of the building is not their pressing problem. Getting paid is.
Construction disputes are document disputes
The Construction Industry Payment and Adjudication Act 2012, in force since 15 April 2014, gives an unpaid party a fast route to an adjudicated decision. The timelines are short and fixed. A party who can put the contract, the payment claim, the supporting records and the correspondence in order within days is in a strong position. A party whose records are scattered across site diaries, WhatsApp and three people’s email is not.
The same is true day to day. Variation orders agreed verbally on site and never priced. Progress claims that take a week to compile. Delivery orders that cannot be matched to invoices. Retention sums nobody is tracking.
This is where AI and automation pay first in construction, because the raw material already exists. It is just unstructured.
Four document jobs worth automating
Reading the incoming paper. Delivery orders, supplier invoices, test certificates and site instructions arrive as scans and photos. A model can extract supplier, items, quantities, dates and amounts for a person to confirm. We run this pattern in production for other industries.
Matching and reconciling. Purchase order against delivery order against invoice. Software does three-way matching instantly once the fields are captured, and surfaces the exceptions.
Assembling the claim. A progress claim is a document assembled from measured work, variations and prior payments. Template-driven generation, which we have built for court documents in a banking litigation platform, turns a week of compiling into a review.
Keeping the record. Every instruction, variation and approval recorded with a date and a name, searchable. If a dispute arises, the bundle is a query.
Certificates and traceability
Materials matter in construction, and so does proof of them. We built a quality control platform for a manufacturer of water reticulation products. Each product is pressure tested, and the resulting test certificate carries a QR code that an authority officer or purchaser can scan to confirm it is genuine.
Contractors receive certificates like these for pipes, valves, steel and concrete. Capturing them against the delivery and the location they were installed is a small step that answers a large question at handover.
Where site AI fits
Drone progress surveys, camera-based safety monitoring and schedule prediction all exist and all work on well-run projects. They need two things most small contractors lack: a baseline plan in digital form, and someone whose job is to act on the output. They are worth piloting on a large job. They are not where to begin.
What this means if you run a construction business
- Digitise the commercial trail first. Contract, variations, claims, certificates, payments.
- Capture site records at the source. A phone form with photos beats a diary typed up on Friday.
- Automate assembly, not judgment. Software compiles the claim. Your QS decides what goes in it.
- Then look at BIM and site AI, with real data to feed them.
A contractor who can produce the full payment record for any project in an hour has already beaten most of the industry. No drone required.