Desert Sol Real Estate: 6 Months of AI Operations Across 130+ Properties
Before Dimora, Nino Tutino was sleeping with his phone. 400+ calls a week across 130+ properties in the Coachella Valley — lockbox questions at 11 PM, pool heat requests during Coachella weekend, checkout disputes the morning of a same-day turnover. His team answered what they could. The rest went to voicemail.
Six months later: 10,000+ calls handled automatically. 50,000+ inbox replies drafted. $35K+ in upsell revenue the team never had time to chase. Same portfolio. Same operations team. Zero missed calls.
Six Months. Real Numbers.
Before Dimora
The Coachella Valley is not a normal vacation rental market. January through April: two Coachella festivals, Stagecoach, and a full snowbird season that fills every property for weeks at a stretch. The phone does not stop. Guests calling about lockboxes at 11 PM. Pool heat requests at 7 AM. Late checkout asks the morning of a same-day turnover where the cleaner was already scheduled.
Desert Sol had 130+ properties and a team fielding all of it. Every high-demand weekend meant overtime and missed calls. Every morning meant working through the previous night's message backlog. The inbox never cleared. The upsell revenue — gap nights, late checkouts, early check-ins — was theoretically there, but no one had time to chase it systematically across 130+ calendars.
The ceiling was visible. Each new property added another 3-4 calls per week, another thread of inbox messages, another set of HOA rules no one had time to memorize. Hiring to keep up meant the margin per property shrank every time the portfolio grew.
What they needed was not more people. They needed a system that could handle Coachella weekend volume and a Tuesday afternoon with identical reliability — one that already knew every property's access codes, drafted replies with accurate guest data, and found gap nights before they disappeared.
What Changed, Module by Module
All eight modules are active at Desert Sol. Here is what each one actually changed for the team — in plain terms, from six months of real use.
Voice AI
10,000+ calls- The 11 PM lockout call rings. Dimora picks it up on the first ring — not voicemail, not a callback
- Every property's lockbox code, WiFi password, parking rules, and house rules: answered correctly for all 130+ properties
- HOA-specific rules per property handled without the team having to script every edge case manually
- Coachella weekend: same answer rate as a Tuesday in January — no extra staffing, no degraded response
- Genuine emergencies transfer to the on-call staff automatically — the AI knows when to hand off
- The caller's reservation details load before the greeting ends — the AI already knows who they are
Inbox AI
50,000+ guest replies- Nino opens Airbnb and the reply is already there — drafted in under 10 seconds, ready to review and send
- Seven specialist agents handle different question types: property details, door codes, availability, upsell acceptance, escalations, and more
- Works across Airbnb, VRBO, and email — the right context pulled for each channel automatically
- Every draft reflects live reservation data: the correct guest name, their actual checkout date, the current access code
- One-week review phase: the team edited replies, the AI learned from every correction, then it ran on its own
- Uncertain messages get flagged for human review instead of guessing — escalation goes straight to the PM's inbox
Revenue Engine
$35K+ in upsells- Late checkout offers go out automatically — pricing set per property tier — no manual outreach
- Early check-in windows identified and offered only when the previous guest is already checked out and cleaning is complete
- Single unbooked nights between reservations detected across all 130+ calendars — extension offers sent to both surrounding guests
- The system never sends an offer that would create a cleaning conflict: turnover time is checked before every message
- When a guest says yes, the acceptance is handled and the reservation updated — no one on the team needs to touch it
- 18% of offers accepted. $35K+ in revenue that used to disappear between checkouts.
AI Learning
118 golden examples- When a PM edits an AI reply, that edit becomes a training example — automatically, without manual tagging
- 118 verified examples built from six months of Desert Sol edits, each one teaching the AI something specific
- New messages are matched against the full library: close matches get used directly, near-matches get adapted
- The closer the match, the less the PM has to touch — most accepted replies need zero edits by month three
- Every correction feeds back into the system the next morning — the AI is measurably better each week
- The feedback loop runs overnight: what your team fixed yesterday shapes what the AI drafts tomorrow
Payment Audit
Daily scans- Every morning at 9 AM, every active reservation scanned for an outstanding balance — no one has to remember to check
- Up to four follow-up messages sent automatically before the case escalates to the team
- If a balance is still unpaid at check-in, the PM gets alerted before the guest arrives at the door
- A Slack alert fires when a case has hit the follow-up limit with no resolution — so it never quietly falls off the list
- Cancelled reservations excluded automatically — the system stops chasing guests who have already left
- The original balance amount is preserved from the moment it was flagged, so partial payments are tracked accurately
Dashboard
Real-time view- One screen shows everything: calls, inbox drafts, upsell offers, payment flags, cleaning jobs, maintenance tickets
- Voice AI logs every call — how many, outcomes, which escalated, when the peaks hit
- Inbox performance at a glance: which replies went out untouched, which were edited, which needed a human
- Revenue Engine tracks every offer — sent, accepted, pending, declined, and total dollars attributed
- Payment Audit board shows every open balance: when it was flagged, how many follow-ups sent, current status
- Live data, no exports — the PM opens the dashboard and the current state of operations is already there
Cleaning
Auto-scheduled- Every checkout creates a cleaning job automatically — checkout time, next check-in window, and cleaner assignment queued without anyone lifting a finger
- Same-day flips flagged immediately: when the window between checkout and check-in is under 4 hours, the team knows before the cleaner does
- Cleaners accept or decline jobs from their phone — no calls, no texts, no manual coordination
- Completion photos uploaded by the cleaner from the property, visible to the PM before the next guest arrives
- Four views — Calendar, Today, By Cleaner, All Jobs — so the right person sees the right information
- A reconciliation runs every morning at 6 AM to catch any jobs that should exist but were missed overnight
Maintenance
AI work orders- When a guest reports a maintenance issue — by call or message — a work order is created automatically, no manual entry needed
- Contractors see whether the unit is occupied before they show up: no surprises for the guest, no wasted trips
- Emergency issues route to on-call contractors immediately — the system knows the difference between "the WiFi is slow" and "there is water on the floor"
- Preventive schedules run across every property: HVAC filters, pool checks, pest control — from weekly to annual
- Every job has a paper trail: who was assigned, when they accepted, what they found, what they fixed, photos attached
- Nothing falls through. The ticket exists from the moment the guest reports it to the moment the contractor closes it.
Tasks
Cross-module- One board shows every open item: cleaning turnovers, maintenance work orders, and anything the team added manually
- Kanban or list view, filtered by type, source, or property — the PM sees whatever cut of the workload matters right now
- Every task shows where it came from: a guest call, an inbox message, or a manual entry by the team
- Status moves from open to in-progress to complete — visible across the whole team, no separate tracking spreadsheet
- Assigned contractor or staff member visible without switching to another module
- Overdue items surface automatically — nothing goes quietly past-due while buried in a list somewhere
Six Months, Month by Month
Day One: The Phone Goes Quiet
Desert Sol went live across all 130+ Coachella Valley properties in a single day. Voice AI picked up from the first call. No warm-up period, no scripted demo calls — real guests, real questions, answered from the first ring. Nino's team forwarded the main number and watched the call volume drop from their plates.
Nino Reviews Every Draft
Inbox AI started in review mode: every guest message got a draft, every draft went to the team before sending. Nino and his managers read each one. When the AI got it right, they approved. When it missed something, they edited — and every edit became a teaching example. By the end of week two, 80% of common questions were coming back accurate with no changes needed.
Coachella Weekend, Zero Scramble
The first major festival weekend arrived. Historically: all-hands, overtime, calls going to voicemail. This time: Voice AI handled every inbound call. HOA-specific rules, parking restrictions, pool heat procedures — answered correctly for each property without the team fielding a single one. First 100 calls in the books, zero missed.
Upsell Offers Start Arriving
The Revenue Engine identified its first gap nights and late checkout windows. Desert Sol set offer pricing per property tier, and the offers went out automatically through guest messaging. The team stopped chasing revenue manually. Guests started saying yes at 18%.
The Inbox Runs Itself
At 100+ training examples, the Inbox AI reached the point where Nino's team was approving replies without changes most of the time. They switched to autonomous mode. Messages came in, drafts went out, guests got answers in under 10 seconds. The PM team shifted from managing inbox to managing exceptions — the small fraction that needed a human eye.
What the Numbers Look Like Today
10,000+ calls handled without a single missed connection. 50,000+ guest replies sent and sent. $35K+ in upsell revenue that used to go uncaptured. 118 training examples. Daily payment audit running every morning at 9 AM. All 8 modules active across the full 130+ property portfolio. Same operations team as month one.
"Dimora handles guest communications across our 130+ Palm Desert properties. Voice AI answers every call on the first ring. Inbox AI drafts replies in under 10 seconds. The Revenue Engine has generated $35K+ in upsells we would have left on the table. Six months in: 10,000+ calls handled, 50,000+ guest replies, zero missed calls. We are managing more properties with the same team — and the phone stopped waking me up at night."
See what this looks like for your portfolio
Most clients are live within 48 hours. Desert Sol started seeing results on day one. Book a 30-minute call — we will walk through the real call logs and draft data from Desert Sol production and show you what the same setup would look like across your properties.