How We’re Building AI that Puts Property Managers First
Four steps to launching AI users trust, from our journey with Revela AI
What would it take for you to trust AI with your business?
Not as a chatbot. Not as a side tool. But as something that actually takes action: on your data, your tenants, and your owners’ money.
That’s the question Don Renyer, Head of Product, and John DeSilva, CTO, faced when they set out to build Revela AI.
Revela exists to make life easier for property managers. Renyer and DeSilva knew AI could streamline their work even further—it shines at repetitive tasks like updating listings and sending work orders.
But the first step was building an AI tool users could actually trust.
Property managers handle sensitive information, making decisions that affect tenants’ homes and owners’ bottom lines. They’re skeptical about outsourcing those tasks to AI, for good reason.
Using AI in property management is a struggle, especially when it sits on top of fragmented, unreliable data. Without a trusted financial and operational foundation, AI outputs create more risk than value.
But Revela wasn’t starting from zero. The platform already unifies the data property managers rely on; a native, embedded AI tool working across all that data, can provide huge value compared to ChatGPT or the other siloed AI tools on the market.
Building it required a careful approach. So Renyer and DeSilva chose to build collaboratively, with real customers. Revela AI is still evolving, but at every step of the way, it’s been shaped by users' needs.
Today, we’re sharing the four steps of their journey: from AI tools that help users take action to ones that take work off their plate.
- Trustworthy AI starts with listening to users’ needs
- Don’t jump into full automation; start by providing alerts and recommendations before scaling
- Let users determine which tasks are AI-friendly and which are best human-led
People-centric AI starts with listening
DeSilva and Renyer envisioned an AI tool that property managers would rely on, day in and day out. That meant it needed to be trusted, proven, and grounded in how they actually work.
“Polish comes from real-world usage,” Renyer explains. “We’d rather ship a small, safe experience and learn how managers interact with it, than spend months building for assumed behaviors.” — DON RENYER, dIRECTOR OF PRODUCT AT REVELA
The plan was to build in public, iterating based on feedback and usage from Revela’s actual users. This would mean earning trust before building assumed automations, and learning what customers want from an AI tool.
“AI is a newer technology for everyone. It’s not something you build behind closed doors,” Renyer continues. The message is less, “try this chatbot,” more “we’re building this - do you want to be part of it?”
Four stages to high-trust, high-value AI support
The Revela team wanted to listen to users’ actions, not just their words. "We chose to launch intentionally small,” says Renyer. “Our roadmap is shaped by real behavior, showing us where property managers get the most value.”
What were property managers jumping at the chance to outsource? Which processes take 20+ manual clicks, but flow the same way 90% of the time?
To gather those insights, Renyer and DeSilva decided to start with a controlled Beta. Based on this feedback, Revela AI would evolve in four stages:
- Alerting users when an anomaly is detected, and action needs to be taken
- Making recommendations helping with steps of the action
- Fully automated workflows
- Stable, clear guidelines for where AI shines
Crucially, data from each phase would inform the next. In short, how property managers used the AI assistant would expand what it could be used for.
Stage 1: Alerting users to potential problems
At launch, Revela AI flags irregularities that need looking into. The AI is there to analyze patterns, behavior, and data anomalies, then prompt the user to take action themselves.
Data-gathering starts. Which alerts are most useful to managers? Which outliers do they appreciate being informed about, and use to take concrete action?
- In action: Alerting when a tenant is behind on rent, or a vendor is overloaded with assigned maintenance tickets.
Stage 2: Recommending and enabling action
Based on user behavior, Revela identifies the most popular and useful alerts. Then, it suggests next steps to take, and helps prepare parts of the workflow.
Data-gathering continues. What recommendations and suggestions do users accept? These will inform the next step: building automations.
- In action: Recommending an email to the overdue tenant, and drafting a message asking when they’ll be able to pay. Or recommending an alternative vendor because the one selected has a high volume of open maintenance tickets.
Stage 3: Controlled automation for tasks users trust
At this stage, usage has revealed the low-risk, routine workflows that property managers trust AI to help with. These are perfect opportunities to save users’ time with full automation. All automations are transparent and often easily reversible, maintaining the audit trail and keeping final control with the user.
Here, data-gathering helps Revela identify which areas of operations are the best fit for AI. For example, do property managers prefer to automate tenant communications, vendor management, or both?
- In action: Emailing the overdue tenant to check on their financial situation. Creating and assigning out routine work orders for maintenance.
Stage 4: Develop guidelines for AI-vs human-led tasks
Some tasks should always keep a human in the loop. Revela looks to their users to show them which ones. Revela AI’s workflows were built on the foundation of contextual data. Over time, even more user data will keep making them more helpful—and more grounded in how property managers work.
These findings can be crystallized into criteria about when to use (and not use) AI. If and when automation is introduced to some higher-stakes workflows, it must always informed by users’ real preferences and behavior.
- In action: Determining that paying routine vendor invoices can be automated, but evicting a tenant must always be done by a human.
Towards a 24/7, always-on AI PMC assistant
By listening to their users, Revela is learning what property managers want from AI.
"We're seeing exactly what we hoped for, and a few things we didn't expect. Property managers start by asking broad discovery questions to learn what their new AI assistant is capable of. Then, they quickly shift into real work, such as updating listings and creating work orders."— DON RENYER
Renyer and DeSilva have built for real people and their roles, not internal assumptions. That’s reduced risk, revealing what’s AI-friendly and what’s best human-led. “From this small, controlled beta group, we’re learning things we couldn’t have from testing ourselves,” says Renyer.
On the ground, these aren’t abstract efficiency gains.
Integrating AI into property management workflows only works when it’s built on clean, unified financial and operational data. AI will work best when introduced in stages, mindful of workflows and user behavior.
With Revela AI, the goal is to make life better for property managers, tenants, and owners. A people-first approach is the best way to get there.
