Published 2026-09-30 · Last updated 2026-10-28 · 5 min read

AI Maintenance Triage Workflow (2026): Turn Tenant Messages into Prioritized Work Orders

Maintenance requests arrive the way tenants talk: a 11:40 PM text that says "water under sink help," a voicemail about a funny smell, an email with three issues and no unit number. The landlord's real job is triage — figuring out what is urgent, what it probably is, who should fix it, and what to tell the tenant — and it usually happens in the cracks between everything else.

The AI-native version turns that triage into a two-minute structured pass per request. You paste the tenant's message, the AI converts it into a work order with an urgency score and a draft reply, and you approve or adjust. Nothing about the actual repair changes; the chaos around it does.

Step 0: Define your urgency rubric (10 minutes, once)

The AI can only score urgency against a rubric you write. Here is a starting point — adapt it to your properties:

• P1 — Emergency (respond within hours): no heat in winter, no water, sewage backup, gas smell, major active leak, lockout, anything making the unit unsafe or uninhabitable.
• P2 — Urgent (24-48 hours): appliance failure (fridge, stove), partial plumbing/electrical failure, broken exterior lock, pest evidence.
• P3 — Standard (within a week): dripping faucet, cosmetic damage, slow drain, minor appliance issues.
• P4 — Cosmetic / next turnover: scuffs, caulk, touch-up paint, non-functional cosmetic fixtures.

Write your version down, including your actual emergency list. This rubric goes into the prompt every time, and it is also worth sharing with tenants so their "emergency" and yours mean the same thing.

Step 1: Convert each message to a work order (Prompt 1)

One chat per request keeps things clean. Paste the tenant's message verbatim, then:

```

You are helping a landlord triage a maintenance request. Convert the tenant's message below into a structured work order.

Tenant's message: [PASTE VERBATIM]

Unit: [UNIT] | Date received: [DATE] | Channel: [text/email/call notes]

My urgency rubric:

[PASTE YOUR RUBRIC]

Output exactly these fields:

1. Issue summary (one sentence, in plain terms)

2. Urgency: P1/P2/P3/P4 with a one-line reason citing my rubric

3. Safety or habitability flag: YES/NO — if YES, quote the part of the message that triggered it

4. Likely trade: plumber / electrician / HVAC / handyman / appliance tech / landlord visit first

5. Questions I should ask the tenant before dispatching (max 3, only if the message is genuinely ambiguous)

6. Draft tenant reply (under 60 words): acknowledge, state the urgency level in plain language ("this counts as urgent — I'll have someone out within 48 hours"), and ask the follow-up questions if any. Sign "- [Your Name]"

Rules:

• If the message suggests a P1 emergency, put "CONFIRM EMERGENCY — call tenant now" at the very top, before anything else.
• Do not diagnose beyond the trade assignment. No repair instructions, no cost estimates.
• Do not invent details the tenant didn't provide.

```

The draft reply is the quiet win here. Most landlords either reply instantly with something vague ("I'll look into it") or go silent for two days while they figure out the plan. A specific reply — acknowledging the issue, naming the timeline, asking one clarifying question — buys you the scheduling time without the tenant feeling ignored.

Step 2: Batch the week's requests (Prompt 2, Monday morning)

If requests trickle in all week, do one batch triage pass every Monday:

```

Here are this week's maintenance work orders (pasted from my triage chats):

[PASTE THE STRUCTURED WORK ORDERS]

Build my week's maintenance plan:

1. Order all open items by urgency (P1 first), then by date received.

2. Group by trade so I can batch vendor calls: all plumbing together, all handyman together.

3. For each item: one-line status (waiting on tenant answer / ready to dispatch / scheduled), and the next action with an owner (me / vendor / tenant).

4. Flag anything open longer than its rubric timeline — those are my overdue items.

Do not close or resolve anything. This is a planning view only.

```

Fifteen minutes on Monday replaces the mental background process of "am I forgetting a maintenance thing?" that otherwise runs all week.

Once the weekly plan exists, track every item to completion with the Maintenance Work Order Log — one row per work order with status, vendor, and cost, so nothing overdue slips through the cracks.

Guardrails

Emergencies bypass the system. If a tenant calls about gas, flooding, or no heat in winter, you do not paste it into a chat first — you act, then document. The prompt's "CONFIRM EMERGENCY" line exists for the ambiguous cases (the 11:40 PM "water under sink"), not the obvious ones.

Habitability issues are legal matters, not just maintenance. If the AI flags a habitability issue, your state's warranty-of-habitability rules set the timeline, not your rubric. Know your state's required response window and put it in the rubric where it belongs.

Landlord vs. tenant responsibility. Before dispatching, check the lease: clogged garbage disposal from misuse, broken blinds, and similar tenant-caused issues may be billable to the tenant under your lease terms. Add a "per lease section X, tenant-responsible" line to the work order when it applies — the AI can include the field, but only you know the lease.

Keep the paper trail. Save each work order: date received, urgency assigned, vendor dispatched, date resolved, cost. When a tenant later claims "you never fix anything," the log answers. When you evaluate vendors at year-end, the log tells you who was fast and who was not.

Related reading: Landlord Maintenance Request System 2026 · Landlord Emergency Maintenance Playbook 2026

Dispatching the vendor (Prompt 3)

The work order is triaged; now someone has to tell the plumber what is actually wrong. Vague dispatch messages ("tenant has a leak, can you check?") cause vague scheduling and return visits. Send the structured version instead:

```

Write a vendor dispatch message from this work order:

[PASTE THE STRUCTURED WORK ORDER]

Include: unit and access details [paste: lockbox code / tenant contact / access window], the issue summary, the urgency level and what it means for scheduling, and one question for the vendor (their earliest availability).

Keep it under 100 words, professional, no fluff. End with "Please confirm receipt and your earliest window. - [Your Name]".

Rules: do not include the tenant's personal contact details beyond what's needed for access. Do not authorize a spending limit I didn't state — if I want a cap, I'll add: "Do not exceed $[X] without calling me."

```

The spending-cap line deserves emphasis: never let a dispatch message imply open-ended authorization. State the cap or state that you want a call first. Every time.

Spotting the repeat offenders (monthly, 10 minutes)

Individual work orders hide patterns. Once a month, paste the month's log into a fresh chat:

```

This is my maintenance log for [month]: [paste: date | unit | issue | trade | cost | days to resolve, one per line]

Analyze for patterns:

1. Any unit with 3+ requests this month — list it and the issues. Flag if the issues look related (e.g., repeated plumbing calls in the same bathroom).

2. Any issue type recurring across units (same appliance model failing, same pest, same leak pattern).

3. Any vendor with average days-to-resolve 2x the others.

4. Total spend by trade and by unit, highest first.

For each pattern found, suggest one preventive action I could take (e.g., "replace rather than repair," "schedule preventive visit," "talk to vendor about response time"). One line each — I'll decide.

```

This is where the workflow pays for itself twice. The third plumbing call in the same bathroom is not three maintenance events; it is one replacement decision you should have made two calls ago. The monthly pattern pass turns your log from a record of the past into the thing that prevents next month's emergencies.

Frequently Asked Questions

What are the four urgency levels in the triage rubric?

P1 — Emergency, respond within hours: no heat in winter, no water, sewage backup, gas smell, major active leak, lockout, or anything making the unit unsafe or uninhabitable. P2 — Urgent, 24–48 hours: appliance failure, partial plumbing or electrical failure, broken exterior lock, pest evidence. P3 — Standard, within a week: dripping faucet, cosmetic damage, slow drain, minor appliance issues. P4 — Cosmetic / next turnover: scuffs, caulk, touch-up paint, non-functional cosmetic fixtures.

What does the triage prompt turn a tenant message into?

A structured work order with exactly these fields: a one-sentence issue summary, a P1/P2/P3/P4 urgency rating with a one-line reason citing your rubric, a safety or habitability flag (quoting the triggering words if yes), the likely trade, up to three follow-up questions only if the message is genuinely ambiguous, and a draft tenant reply under 60 words. If the message suggests a P1 emergency, "CONFIRM EMERGENCY — call tenant now" goes at the very top.

Do emergencies ever bypass this workflow?

Yes. If a tenant calls about gas, flooding, or no heat in winter, you act first and document later — the workflow is for the ambiguous cases, not the obvious ones. Habitability issues are also legal matters: your state's warranty-of-habitability rules set the response timeline, not your rubric, so put that required window in the rubric itself.

How does the monthly pattern pass spot repeat problems?

Paste the month's log (date, unit, issue, trade, cost, days to resolve) and look for any unit with 3+ requests, any issue type recurring across units, any vendor averaging twice the days-to-resolve of the others, and total spend by trade and by unit. The point is that repeated calls are often one decision, not separate events — the third plumbing call in the same bathroom is a replacement decision, not three maintenance events.