The Data Foundation Required for Meaningful AI Implementation

John DeSilva co-founded Revela nearly ten years ago and architected the platform from inception. Don Renyer brings product leadership experience from multiple industries to property management technology. Together, they've guided Revela's AI implementation—a deliberate approach that prioritizes user trust and operational context over feature velocity.

They know what separates effective AI deployment from failed experiments: property management companies need deeply integrated systems with unified data foundations, not another standalone tool bolted onto existing workflows.

Key Takeaways

• Embedded AI systems with a complete operational context deliver better decisions than general-purpose tools requiring manual context setup.
• Trust-building through phased capability expansion (alerts, recommendations, automation) prevents user resistance.
• Unified data foundations enable AI to consider the full operational picture
• Natural language interfaces remove learning curves while a permission-based architecture maintains security
• AI shifts property managers from task execution toward strategic asset management

The Problem With Bolt-On AI Solutions

Property managers already juggle multiple systems. When they add general-purpose AI tools like ChatGPT to their workflow, they create another context-switching burden.

"You lack the context to go from one platform to another, and then of course, it can't actually take action either. It can give you a listing description, but it can't update the listing description. You'd have to copy and paste it, make any formatting or other changes directly within your platform."_{— John desilva, Chief architect AT REVELA}

The limitation extends beyond convenience. A property manager using ChatGPT to improve a listing description must manually provide the address, square footage, nearby amenities, market rate comparisons, and property history. Even after investing that effort, they still need to copy the result back into their property management system and format it correctly.

Renyer describes working with property managers who use standalone AI tools today: "They generate market listings and then ask, 'Can you make this sound better? Because I'm not a marketing person.' Those are all extra steps that a property manager has to take."

Revela's embedded AI handles this differently. When property managers ask the AI to update a listing, it accesses the platform's unified data, connects to third-party integrations, and pulls web information to populate features, amenities, square footage, suggested market rates, and nearby attractions. This all happens without manual input. The AI delivers complete, market-competitive listings in seconds rather than requiring property managers to gather information from multiple sources.

The core issue remains data access. General-purpose AI lacks visibility into your accounting records, maintenance history, vendor performance, owner preferences, and tenant communications. Every task requires rebuilding context that should already exist.

The Embedded AI Architecture

Property management operations span multiple domains that constantly intersect. Renyer describes a common scenario: "I'm a property manager, and I'm running some owner balance reports and working on some end-of-month financials. I get a phone call that a resident's sink is leaking." This is where Revela AI comes in.

Without interrupting their workflow, they simply tell the embedded AI: "I've got a leaky sink at 123 Main Street." The system already knows that the address exists in their portfolio. It understands water damage spreads quickly. It can access vendor records showing which plumbers complete jobs reliably in that area. It creates the work order and dispatches the vendor while the property manager continues their financial work.

"Think of it as almost another employee. You can ask them to jump on it and have the confidence that they have the competence to take care of that for you while you handle the task at hand."_{— DON RENYER, dIRECTOR OF PRODUCT AT REVELA}

This capability requires three architectural foundations that differentiate embedded AI from standalone tools:

Unified Data Foundation

"For somebody that's using a third-party maintenance tool that's different from their property management software or their core system,” DeSilva adds. “An AI tool is able to connect to that third party via API or even a direct integration, but it's unlikely that the full picture will be there."

The AI might know vendor names and specialties, but lacks performance ratings and availability. Those gaps produce suboptimal recommendations that a property manager must then manually verify and correct.

"AI is only as strong as the data," Renyer notes. "Revela already unifies accounting, leasing, maintenance, communication, and vendor management into one ecosystem. That gives our AI complete operational context."

When property managers ask, "Who do I owe money to?" the AI doesn't need additional context—it already functions as a property manager accessing a complete operational dataset.

Permission-Based Architecture

Embedded AI inherits user permissions instead of operating with system-wide access. A maintenance supervisor can create work orders but cannot update listings or process owner disbursements. An administrator has broader capabilities.

"We made it have parity with the users’ permissions that are engaging with the AI," Renyer explains. "If I am just the maintenance supervisor with a property management company, I'm not going to be able to utilize Revela AI to update a listing."_{— DON RENYER}

This approach addresses both security and trust. Property management companies handle sensitive financial data and tenant information; an AI that can access everything creates unacceptable risk. Permission parity also means every action is logged alongside the specific user who initiated it, maintaining accountability.

Natural Language Interface

Text-based interaction removes the need to learn complex prompting techniques or navigate through multiple system sections. Property managers already know how to communicate since embedded AI speaks their language.

DeSilva describes the UI advantage:

"Property management is a very interconnected web. That leaking faucet has a financial impact. It has a resident satisfaction impact. You have to communicate with a vendor. You have to understand what other work orders are open. A text-based interface is quite a natural way to bring all those things together in one flow where you don't have to jump around."_{— John desilva}

Building Trust Through Phased Capability Expansion

Even well-designed AI faces adoption barriers. News coverage emphasizes failures and risks, making users naturally apprehensive. Revela addresses this through graduated capability deployment.

Renyer describes their three-phase approach using overdue rent collection as an example:

Phase One: Alerts Only

The AI identifies exceptions without taking action. "In the initial phase, it will find that a tenant is $3,000 and sixty days behind on rent. You should look into this." Users gain confidence in the AI's analytical accuracy without risking incorrect automated actions.

Phase Two: Recommendations With Approval

After users trust the AI's problem identification, it begins suggesting solutions. After it finds a tenant behind on rent, you’ll receive an outline of the next steps. "For example, the AI recommends that you send an email communication to John asking him when he's gonna be able to pay." The property manager approves or rejects the recommendation. Approving the recommendation grants the AI permission to execute that action immediately, so the communication gets drafted and sent. Consistent approval rates for specific recommendation types signal readiness to advance those workflows to full automation.

Phase Three: Controlled Automation

For proven workflows with consistently high approval rates, the AI executes automatically while notifying users after the fact. "We went ahead and drafted the communication following up on that. We've popped the flag on his account. There's nothing that the property manager has to do except receive a notification that that action occurred."

Critically, not every workflow advances to automation. "There are certain elements within our system that we may never fully automate," Renyer notes. "We may always want eyes on this particular type of action."

The Data Consolidation Imperative

Over the next few years, property management companies face a choice: consolidate operations onto unified platforms that enable embedded AI, or continue managing fragmented systems that limit AI effectiveness.

"All of these complicated, complex manual workflows will be automated," Renyer predicts. "Instead of having to go do actions in a multistep process, you'll be receiving an alert that an action was taken. Here are the highlights that you need to know."_{— DON RENYER}

Companies with unified data foundations and embedded AI will operate with people able to do higher-value work. Their competitors using standalone AI tools will still manually bridge information gaps between systems.

The reality of integrating these workflows supports team growth. Each person can support more units as an individual, and leverage AI for faster daily tasks. At the same time, the repeating cycles of month-end closing, leasing, maintenance, and others will run more smoothly. Owner satisfaction rises, and a property management company can successfully scale while improving margins. 

The window for building this foundation closes as market expectations shift. Tenants increasingly expect modern digital experiences. Owners want sophisticated reporting and proactive asset management. Regulatory requirements grow more complex. Property managers who can't deliver because they're buried in operational tasks will lose business to those who've automated the routine work.

DeSilva emphasizes data completeness:

"There's a lot of data out there. We collect a lot of data points about historical interactions with vendors, historical financial information, and time to complete things. But who's making good use of that? This is a very natural way to surface those data points and make use of them."_{— John desilva}

The companies that win will be the ones who built the data foundations and embedded architectures that let AI actually work.

Property management companies exploring embedded AI can see these architectural principles in action through Revela's AI beta program. The platform demonstrates how unified data foundations, permission-based access, and natural language interfaces work together to eliminate context-switching and enable genuine automation. Property managers interested in experiencing AI that operates within their existing workflows, rather than adding another tool to manage, can learn more about current capabilities and beta access at revela.co/ai-beta.

‍