The Northeast Monsoon was expected to begin on 27 October and run until March. MetMalaysia expects three to five spells of continuous heavy rain between November and January, with flood risk in low-lying areas and near rivers.
The same week, in a written reply to Parliament, the Digital Ministry said not every one of the country's 156 local councils needs its own AI system. Its plan is to build once and reuse: a shared data platform, a bank of AI algorithms and a marketplace councils can draw from. So far the AI City projects people hear about are mostly about traffic.
For the next three months, the more useful shared piece may be far less glamorous: what happens between a resident's flood report and a crew clearing the drain.
The expensive problem: one drain, many tickets
When heavy rain hits, reports arrive everywhere at once: the council's complaint app, the hotline, WhatsApp messages to a councillor, and public posts tagging the council. Recent flash floods in Selangor show the pattern. In one case, a resident appealed to the minister on Threads, and the council's reply had to ask for the full address before anyone could inspect. In Sepang, a councillor said drains due for their second yearly clearing had not been cleared when water entered about ten homes on 29 October.
- the same blocked drain becomes several tickets, sometimes with different owners
- a photo without a precise location means a call-back, or a crew sent to the wrong spot
- nobody can quickly see the drain was due for clearing, or that roadworks nearby hold a permit
Each of these is a handoff gap, not a shortage of effort. Officers spend the storm re-keying, calling back and forwarding, while residents see silence and post again.
The workflow: from report to closed case
Here is how AI fits, one step at a time. AI reads, groups and checks. Officers keep every decision about priority, crews, contractors and closing a case. Councils have already digitised a lot, with complaint platforms such as iTegur and PBTCare, and the housing ministry is adding AI to flag incomplete development applications. This sits on top of the complaint system a council already runs, rather than replacing it.
- 01CaptureAI does
- 02Merge & locateAI does
- 03Suggest routeAI does
- 04Prioritise & dispatchPerson decides
- 05Check closureAI does
- 06Close & replyPerson decides
1. Capture every report in one queue
Residents keep using the channels they already use. Each report, whether app, hotline note, WhatsApp or a public post, lands in one queue with its photos, time and any location shared. Nobody has to retype a screenshot.
2. Merge duplicates and pin the location
AI reads the text and photos, works out the most likely location from pins, street names and landmarks, and groups reports about the same spot into one case. Anything it is unsure of is marked for a person to check, with the original messages one click away.
- Location
- Jalan Permai 3/2, culvert ✓
- Issue
- Blocked drain, overflow ✓
- Homes affected
- About 6 (residents say) Check
- Cause noted
- Construction debris Check
- Duplicates
- 6 merged into one case ✓
3. Check the registers and suggest an owner
Plain rules compare the case with what the council already knows: who maintains the drain, when it was last cleared, and whether active works permits sit nearby. The output is a suggested owner and a short list of flags, not a decision.
| Check | Register says | Result |
|---|---|---|
| Drain asset | MD-0412, council drain | ✓ |
| Last cleared | May (due twice a year) | Flag |
| Nearby works | Road permit RW-0931 active | Flag |
| Suggested owner | Engineering, drainage | ✓ |
4. The duty officer sets priority and dispatches
One named officer sees the case, the evidence and the suggestions together. They decide the priority, send the crew to the exact spot and decide whether the permit holder gets a notice. Their decision is recorded against the case.
- Water reported entering about 6 homes: suggest top priority
- Debris may come from works under permit RW-0931: notify the contractor
5. Check the closure evidence
When the crew finishes, AI checks that the after photo matches the location and the job, that the remarks are complete, and drafts one plain update for everyone who reported it. Gaps are flagged before anyone presses close.
- ✓Before and after photos at the same location
- ✓Crew remark: culvert cleared, debris removed
- CheckContractor notice issued, reply pending
- ✓Update drafted for all 7 reporters
6. The engineer closes it and residents hear back
The drainage engineer reviews the evidence and closes the case, or keeps it open for a re-inspection after the next heavy rain. Every resident who reported it gets the same update, so nobody has to post again to find out.
- 29 Oct, 7:12 PMFirst report; 6 more merged
- 7:40 PMRouted to drainage, flags attached
- 8:05 PMFarah dispatches crew
- 30 Oct, 2:30 PMEngineer closes; 7 reporters updated
What to measure before the next spell of rain
Before buying anything, including from us, measure the current process in one zone across the first few rain spells. The first number is the one residents feel. The rest explain why it is high.
If the numbers are already good, AI is not the priority. If they are not, the council has a case it can explain to its state, its auditors and its residents.
- 01Hours from first report to crew on sitemedian, by priority: the headline number
- 02Reports per incidenthow many tickets one blocked drain creates
- 03Reports needing a call-backshare missing a usable location
- 04Reopened casessame spot within 30 days
- 05Cases closed without photo evidencecount per month
Where these projects go wrong
Most failures here are not about the AI model. They come from asking residents to change their habits in the middle of a storm, or from letting software make a call that affects people's homes and a contractor's liability.
Residents' reports contain personal data, so keep them inside the council's own systems with clear access rules. And this is exactly the kind of pattern the ministry's build-once approach suits: the intake, merging and routing logic is much the same in every council, even if the drains are not.
A practical first step
Pick the zone that floods most. Sit with the complaints unit, the duty officers and the drainage team to map how a report really moves. Baseline the numbers above while the rain is still coming. Then run a bounded pilot with a clear stop rule: if crews do not reach sites faster, stop and keep what you learned.
That is a better first AI project for a council than another dashboard nobody acts on.
Tenantra helps Malaysian organisations, public and private, map one costly workflow, measure it and build the software to fix it, starting small. If monsoon complaints swamp your team, tell us about it.
This article describes an illustrative workflow, not a client case study. Names, cases, places and figures are invented for illustration. It is general information, not legal advice.
- Weeks 1–2Map how one zone's flood and drain reports flow today
- Weeks 2–4Baseline the five numbers during the rain spells
- Weeks 5–14Pilot one queue and routing in that zone
- DecisionTime to site and call-backs down? Extend. If not, stop.
Sources
- Start of Northeast Monsoon expected on Oct 27, MetMalaysia says (The Star, 23 Oct 2026)
- Northeast monsoon set to bring flood risks (The Star, 24 Oct 2026)
- AI City Initiative drives improvements in traffic management (The Edge Malaysia, 26 Oct 2026)
- MBPJ orders contractor to clean drains after Lembah Subang floods (Citizens Journal, 27 Oct 2026)
- After another downpour, Selangor resident asks minister to step in over recurring floods (AsiaOne, 18 Oct 2026)
- Uncleared drains trigger Sepang flash floods (The Star, 30 Oct 2026)
- Digital transformation to boost PBT service delivery (The Star, 13 Sep 2026)