The Story of Xenia

The Sublet That Started Everything

It was August 2021. I was leaving Milan to go back to Cambridge for my final year, and I had a problem: I was paying rent on an apartment I wouldn't be using.

So I did what any broke student would do. I threw it on Airbnb.

Within two weeks, I realized something that changed everything: the rent I was paying was half of what I could make on Airbnb. I wasn't just covering my costs. I was profiting.

When the semester started, I made a decision that seemed ridiculous at the time. I kept paying rent on an apartment 1,500 kilometers away. Not to live in it. To rent it out.

But I had a problem. Someone needed to handle the guests. The check-ins. The cleaning. The chaos.

I started calling my friends. Most said no. Then I called Marco.

He said yes immediately. But more importantly, he asked a question that would define our entire company: "Why do we need to be there at all? Why can't we just put a lockbox on the door and let guests check themselves in?"

That was it. That was the moment.

A lockbox. Self check-in. No human required.

It sounds obvious now. In 2021, it felt like a revelation. Because for the first time, I understood something fundamental: technology could cut out the work. We could manage more properties with the same effort. This thing was scalable.

The Notebook Era

The first apartment we ever onboarded for another owner was in early 2022. Our "system" was a physical notebook. I still have it.

When that owner asked about our other properties, we sent them fake Airbnb links to listings we didn't manage. We were two guys with a notebook pretending to be a company.

It worked. They signed.

We quickly graduated to spreadsheets. Google Sheets became our operating system. We wrote scripts to manage cleaning schedules. We built formulas to track occupancy. We were power users of a tool that was never meant to run a property management company.

This was before we even knew PMSes existed.

The first time we saw a Property Management System, it felt like discovering fire. Finally, something that could handle multi-channel distribution without us manually updating three different calendars. Finally, something built for this industry.

And then we realized: it wasn't enough.

The Decision to Build

By early 2024, we had 30 apartments. We were managing them on nights and weekends, during lunch breaks, between meetings at our day jobs. Every day was the same: check-in instructions copy-pasted at 2am, cleaners coordinated via WhatsApp chaos, guest emergencies handled while pretending to focus on something else.

We weren't running a business. We were running on fumes.

That's when I made a decision. I was going to stop treating property management as a people problem and start treating it as an engineering problem.

I had some experience building rudimentary chatbots and agents. So I started building. A task management system (we didn't know Breezeway existed). An apartment scoring algorithm that ranked properties by urgency so we could prioritize where to send our maintenance team. A dispatch algorithm to optimize routes. An Airbnb chatbot to automate messages.

I was trying to build a system that could coordinate issues end to end. Not just reply to messages about WiFi passwords, but actually orchestrate work and information.

I quickly realized I couldn't build it alone. I wasn't technical enough. I needed a team.

The Host AI Chapter

I reached out to Host AI (now called Conduit), a YC W24 company building exactly what I was trying to build. I still remember the first call with Cole.

I told him I was a huge fan of what they were doing. Also that I was building a competitor. And I was asking for insights.

His response was basically: "Why would I give you information about my competitor?"

Fair point.

Five months later, after talking to every company in the space (Boom, Dharma, Enso Connect, and half the Italian market), I reached out again. This time, not as a competitor. As someone who wanted to join.

They said yes.

Moving to San Francisco was a dream fulfilled. The startup culture. The energy. Building what I'd always had in mind: an AI agent that could handle guest requests end to end. Not just answer FAQs, but coordinate, orchestrate, and execute.

Here's what I learned: getting an AI to reliably answer basic FAQs is brutally hard. I thought it would take a month to build the knowledge base. It took five. But we built the best system in the industry, and the foundational work for true coordination.

In late 2025, Conduit decided to concentrate on other verticals. But the building blocks were there. And I knew what had to happen next.

I left. I came back to Milan. And I started again, more motivated than ever to change this industry from a people business to a software business.

The Xenia Model

Today, Xenia manages 150 apartments in Milan. We generated €3 million in revenue in 2025.

Our team? Four people.

One commercial person. And three operational people who alternate, so there's always one property manager active at any given time. That's it.

We grew the portfolio 5x in two years. Our headcount barely moved.

This isn't luck. It's architecture.

The Philosophy

Most property managers scale by hiring. More apartments means more people for guest comms, more people scheduling cleaners, more people handling maintenance.

We rejected that model entirely.

At Xenia, we follow one rule: if a computer can do it better, a human should never touch it.

Every process follows the same progression:

Tracking → Standardization → Optimization → Automation

You can't automate chaos. You first have to understand it, standardize it, then systematically eliminate the human in the loop.

Our north star? We want to be the Waymo of hospitality. Just as Waymo automates driving, Xenia automates the "driving" of hospitality operations: the coordination, communication, and execution that traditionally requires armies of people.

The Stack

Here's the thing about scaling STRs in Italy: the generic US-focused tools don't work. Compliance is complex. Invoicing rules are brutal. Most global PMS platforms don't handle Italian fiscal requirements at all.

So we built our stack around what actually works:

PMS: Kross Booking. The only PMS designed for the Italian market. Handles reservations, channel management, invoicing, owner portals, AlloggiatiWeb compliance, and fiscal reporting. Everything talks to Kross as the single source of truth.

Guest Messaging: AI-powered (Conduit heritage). Our AI handles 90%+ of guest inquiries without human intervention. It knows every property. Every house rule. Every quirk. The WiFi question at 3am? Handled.

Field Operations: Breezeway + Custom Backend. Breezeway handles task scheduling. But it's a task manager, not an orchestrator. So we built a backend that pulls from Breezeway, cross-references with Kross reservations, and generates intelligent daily planning for our field team.

Pricing: PriceLabs. Dynamic pricing, calibrated monthly. 15-25% revenue uplift compared to static pricing.

Access: Hermes (Custom Built). We built our own smart access system using Shelly devices. Remote intercom unlocking, real-time monitoring, all from one dashboard. When a guest is locked out at 11pm, we tap a button. No runner required.

Accounting: Custom Reconciliation Scripts. Italian invoicing is a nightmare. We built scripts that generate invoices from Kross data, send them to the Sistema di Interscambio (SDI), and reconcile everything with bank statements. What used to take days runs overnight.

What Automation Actually Means

Let me walk you through what happens when a guest reports a broken air conditioner at 2pm in August.

The system receives the message and classifies it instantly: HVAC issue, urgency HIGH (summer, AC is essential), complexity SIMPLE (fixable by our team). It checks our field staff positions via Breezeway, sees that Alessandro is 2km away finishing a task in 45 minutes.

The dispatch algorithm recalculates his entire afternoon. A routine supply restock at an empty apartment? Deprioritized. A pre-check-in inspection for a 6pm arrival? Stays on schedule but shifts 30 minutes. Alessandro gets a reorganized route with the AC repair slotted in, plus an auto-generated briefing: access code, AC unit location, common fixes for this model, parts in inventory.

If Alessandro can't fix it, the system escalates automatically: checks available apartments for relocation, drafts the offer message, prepares the OTA modification request. All before a human reviews the case.

The guest gets proactive updates throughout. "Our technician Alessandro is on his way, ETA 45 minutes." Then: "Alessandro has arrived." Then: "Issue resolved."

Total human decision-making required? One person reviews the AI's proposed path and hits approve.

Every evening at 6pm, the system generates tomorrow's field operations. This isn't a task list. It's multi-variable optimization: urgency scoring, damage coefficients for unresolved issues, recency flags for apartments not inspected recently, route optimization clustered by neighborhood, guest constraint windows. The algorithm outputs who goes where, in what order, with what tasks, carrying what supplies.

When something changes mid-day (urgent issue, cleaner calls in sick, early check-in request), routes recalculate in real-time.

A checkout at 10am triggers automatic cleaning assignment to the nearest available cleaner. They receive access codes, unit-specific instructions, and the next guest's check-in time as a hard deadline. If they run late, the system messages the incoming guest with a delay notice and complimentary late checkout on departure. If they flag quality issues during turnover (stained mattress, broken blinds), those become maintenance tasks that slot into tomorrow's dispatch automatically.

No coordinator playing traffic controller. The system is the coordinator.

The Real Unlock

Here's what most people get wrong about automation: they think it means replacing humans.

Wrong.

Automation means freeing humans to do what humans do best. Solve novel problems. Build relationships. Make judgment calls.

At Xenia, we're not trying to eliminate people. We're trying to eliminate the repetitive, soul-crushing work that burns out every property manager. Our team doesn't copy-paste check-in instructions. They don't manually reconcile invoices. They don't answer the same WiFi question for the 500th time.

Instead, they design better systems, handle the edge cases that matter, and build the relationships with owners and guests that actually drive the business.

We call this the shift from Property Manager 1.0 (the firefighter constantly reacting to chaos) to Property Manager 2.0 (the Experience Engineer who designs systems and supervises execution).

What's Next

Everyone in the industry knows the endgame. Data silos are no longer acceptable. Property managers need unified systems that can handle all information autonomously. The question is who gets there first.

By end of 2026, Xenia will operate on a single AI-governed backend. A digital twin of the entire company. Every conversation, asset, task, issue, and workflow represented in software. The system will execute the full operational stack with human supervision only where required.

There's a gold rush happening right now to build the first AI agent that can handle property management end to end. Not just answer messages. Coordinate. Orchestrate. Execute.

I've been building toward this since that first lockbox in 2021. Since the notebook. Since the spreadsheets. Since Conduit.

The future of property management isn't more people managing more chaos. It's fewer people, better tools, and systems that run themselves.

The Advice

If you're building a property management company and want to avoid the trap of scaling headcount with apartments:

1. Track first. You can't improve what you don't measure.
2. Standardize second. Create repeatable processes before you automate them.
3. Optimize third. Remove unnecessary steps.
4. Automate last. Only when the process is clean and predictable.

And never copy-paste a check-in instruction again.