Skip to main content
Luna Skills

Set up instructions for recurring rental tasks

When a tenant reports no heat, use the No Heat skill to collect the details and identify safety stops. Review the suggested next step before arranging qualified help.

Open maintenancePlans and fees Explore maintenance
  • Written procedures
  • Guided questions
  • Your approval
15Authored procedures included
3-tierEmergency, high, routine
EditableRewrite any, add your own
ScoredBy outcome, not by activity
One no-heat report, with a clear next step
A rental home with a landscaped entrance
Start with the no-heat request.

Confirm the rental, symptoms and any immediate safety concern.

Tenant reports no heat. Here is what actually happens.

The AI matches the ticket to the No Heat skill and works it: thermostat setting and batteries, breaker and furnace switch, filter condition, with photos collected along the way. A gas odor ends everything immediately and escalates as an emergency. Cold snap outside or a vulnerable resident? Emergency. Mild weather after the checks? High priority, suggested to your HVAC vendor. One cold room? Routine.

Diagnostic questions your coordinator would have asked
Safety stops are hard rules, never left to model judgment
The escalation tier sets the work order's urgency for you
Repair tools beside a contractor’s invoice
Keep the repair, invoice and your review together.

Edit any step. Rank the vendors. Keep every version.

A skill is built as steps, not a wall of text: add, reorder, and reword the questions the AI asks and the checks it walks a tenant through, and see the same script previewed as a chat, as a phone call, and as a text message before you save it. Rank as many vendors as you like per skill, set who gets paged when it escalates, and switch off any skill you do not want running. Every edit is kept, so you can compare versions and restore one in a click.

Step editor with a live chat, voice, and SMS preview
Version history with one-click restore
Edit one skill and the rest keep improving on their own
A rental document ready for review
Review the rental’s document before taking the next step.

Not "used 47 times". Resolved without sending anyone out.

Activity counts tell you nothing about whether a procedure earns its keep. Every skill conversation is recorded by how it ended: resolved without a visit, work order raised, handed to a person, or abandoned partway. Each skill shows you its own result, in words, not just numbers. When a script hands eight conversations in ten to a human, the page says so, because that is a problem with the script and not with your tenants.

Outcome recorded per conversation, not per click
The tenant confirms the fix worked, not the AI
Every conversation links back to its ticket
Rental ledger and transactions ready for review
Check the source records before confirming the entry.
Safety first The stops are not suggestions. Gas odor, sparking outlets, sewage backups, flooding a shutoff cannot stop: these conditions end tenant troubleshooting instantly and escalate as emergencies, every time, in every skill.

Skill questions

What is a Luna skill?
A written operating procedure the AI follows for one situation, like no heat or a water leak: which questions to ask, which steps are safe for a tenant to try, which photos to collect, when to stop for safety, and whether the ticket escalates as an emergency, high priority, or routine.
Can we change what the AI says and does?
Yes. Each of the 15 shipped skills is a set of steps you can add to, reorder, and reword, with a live preview of how it reads as a chat, a phone call, and a text message. Every skill has an on-off switch, you can write custom ones for situations specific to your buildings, and every edit is versioned so you can restore an earlier one. The safety stops are the exception: those are locked and cannot be edited or removed.
If we edit a skill, do we stop getting Rentari's improvements?
Only the parts you actually changed stay yours. Everything you have not touched keeps receiving our improvements automatically, so editing one question in the no-heat script does not freeze the rest of it. You can also reset any skill back to the Rentari version in one click, and your edits stay in the history.
How do we know a skill is actually working?
Every conversation is recorded by how it ended: resolved without sending anyone out, work order created, handed to a person, or abandoned. The headline is resolved without dispatch, because that is the one that saves you a call-out. The tenant confirms the fix worked, not the AI.
Do we need a separate skill for chat, phone, and text?
The same one. A skill is authored once and rendered for each channel: full sentences in chat, spoken lines with no formatting on a phone call, and short segments over SMS. You do not maintain three versions of the same procedure.
Does a skill dispatch a vendor by itself?
A skill can carry a default vendor so the dispatch suggestion is pre-filled the moment the ticket is triaged. Dispatch itself still follows your approval settings; nothing is sent to a vendor without the control level you have chosen.
What happens in an emergency?
Safety stops are hard-coded into the procedures: gas odor, active electrical hazards, flooding that a shutoff cannot stop, and similar conditions end tenant troubleshooting immediately and escalate as emergencies. The AI never talks a tenant through something unsafe.

Give your maintenance line a playbook.

Fifteen procedures are ready the moment you sign up. Rewrite them into your own by Friday.

Try it free For PM companies

Included with AI maintenance triage. No per-ticket fee.