Where AI Actually Saves Time for CRE Owners

AI is everywhere in commercial real estate right now, but the practical question for owners is still the same one it has always been: where is the work breaking down, what is it costing the business, and can the team prove that a technology decision improves performance? In this episode of Peak Property Performance®, Bill Douglas talks with Greg Achenbach, VP of Product at PredictAP, about one of the least glamorous and most useful places to apply automation: accounts payable. The conversation is not about chasing AI. It is about turning repetitive, high-volume CRE work into measurable time savings, fewer bottlenecks, and better operational control. You can listen to the full episode.

Commercial real estate accounting is messy because the building is messy

Greg’s starting point is simple: commercial real estate produces a massive amount of financial activity. Every property generates purchases, repairs, services, maintenance invoices, recurring vendor bills, one-off site-level expenses, and exceptions that need human attention. A single office asset or multifamily property can produce enough activity to strain a small accounting team. A portfolio multiplies the problem.

That matters because accounts payable is not just a back-office workflow. It is connected to property operations, tenant experience, vendor relationships, and risk. If a critical invoice is missed, delayed, miscoded, or routed incorrectly, the consequence can show up as a service disruption, a strained vendor relationship, an avoidable late fee, or a property team spending hours chasing paperwork instead of managing the asset.

Greg described PredictAP as “an AI invoice ingestion and coding platform for real estate companies,” but the more important point was why AP is such a practical automation target. It has volume. It has repeated patterns. It has structured business consequences. It also has enough complexity that simple document scanning does not solve the problem. OCR might read an invoice, but it does not necessarily understand how that invoice should be coded, routed, reviewed, or connected back to the property’s operating reality.

“There’s just a lot of purchasing, a lot of repairing, a lot of services. There’s just a lot of financial activity that happens at the real estate world.”

That is the kind of work where technology can help, but only if the product is built around the actual operating job. The property team does not need an AI label. The accounting team needs fewer low-value touches, cleaner coding, faster review cycles, and better visibility into what is happening across the portfolio.

Do not buy AI. Find the bottleneck first.

Bill pressed on the point that many CRE teams are seeing AI everywhere at conferences and in vendor messaging. “Your AI partner” and “CRE AI” have become common labels. The danger is that the label can become the buying reason. An owner can end up purchasing a point solution that claims AI capability without first identifying the business problem, the workflow bottleneck, or the measurable outcome.

Greg answered from a product builder’s perspective. AI is not one thing. It is a broad marketing term for many types of advanced technology, including pattern recognition, machine learning, natural language processing, and conversational interfaces. A user might think of AI as a chatbot because that is the most familiar consumer experience. But in an AP workflow, the useful capability may be invoice recognition, coding prediction, routing logic, exception detection, or pattern analysis across historical records.

The useful question is not, “Where can we use AI?” The useful question is, “Where are people spending time on lower-value work that keeps them from higher-value work?” In accounts payable, that might mean manually reviewing invoices that follow known patterns, re-entering data that already exists somewhere else, checking coding decisions that should be predictable, or searching across records to find late fees, duplicate charges, or repeated vendor issues.

“No one wants to buy another product. What they want to do is have a job.”

Greg connected this to the “Jobs to Be Done” idea from product strategy. The classic example is that someone does not really need a drill, they need a hole. In CRE, an owner does not need AI as an object. The owner needs a workflow that reduces wasted time, improves accuracy, protects sensitive information, and produces a result the business can measure.

Measurable ROI starts with a specific operating pain

One of the strongest parts of the conversation was the distinction between automation that is possible and automation that is worth doing. Bill noted that OpticWise has seen this pattern in data & digital infrastructure work as well: many things can be automated, but not all of them produce a return. Automation for its own sake can create more expense, more complexity, and more systems for property teams to manage.

That distinction matters for owners and asset managers because AI can become expensive quickly. The cost may show up in software subscriptions, implementation time, integrations, data preparation, staff training, security review, and long-term vendor dependence. If the use case is vague, the owner may not know whether the tool improved performance or simply added another layer to the operating stack.

Greg’s recommendation was to begin with a specific pain point where the impact can be tracked. In AP, that could mean invoice processing time, coding accuracy, exception volume, late fees, duplicate invoices, or the number of manual touches required before an invoice is approved. The same principle applies across CRE operations. If a team wants to use AI or automation in maintenance, utilities optimization, leasing, insurance documentation, or capital planning, the first step is to define the operational bottleneck and the metric that will prove progress.

This is where CRE has a major advantage if owners can get control of their data & digital infrastructure. Real estate companies already sit on mountains of operational and financial data. The problem is that much of it is trapped in disconnected systems, vendor portals, spreadsheets, property management systems, building systems, invoices, service records, and site-level practices. When the data is not governed, portable, or trustworthy, even a strong AI tool can only work with fragments.

• Which invoices include late fees, and why are they recurring?
• Which vendors create the most exceptions across the portfolio?
• Which properties require the most manual AP intervention?
• Which expenses are being coded inconsistently from asset to asset?
• Which operating patterns are invisible because the data lives in separate systems?

Those are not abstract AI questions. They are asset management questions. They affect staff capacity, operating expense control, vendor accountability, and the quality of decisions across the portfolio. When data is accessible and governed, technology can help answer questions that would otherwise require hours of manual review. When data is fragmented and vendor-controlled, the same questions become slow, expensive, or impossible to answer consistently.

The right AI question is a workflow question

Greg’s answer to the AI mandate was useful because it brought the conversation back to work. AI is not one thing. It is a broad label applied to a set of technologies that include machine learning, natural language processing, pattern recognition, automation, and increasingly conversational interfaces. That breadth makes the term powerful in marketing, but imprecise in operations.

For a CRE owner, the practical question is not whether a product uses AI. The practical question is whether the product improves a workflow that is already costing the business time, money, accuracy, or control. In accounts payable, that workflow is easy to see. Invoices arrive from many vendors, across many properties, with different coding rules, approval paths, exceptions, and consequences if something is missed. That gives the technology a real job to do.

“No one wants to buy another product. What they want to do is have a job.”

That “job” framing is especially important in commercial real estate because property operations create messy, local context. A general-purpose tool may be impressive, but it may not understand the difference between a recurring contract expense, a one-time repair, a tenant-billable item, a capital project cost, or an invoice that should be routed differently because of the asset, vendor, GL code, or ownership structure. In CRE, those distinctions matter because they affect reporting, approvals, recoveries, budgeting, and asset-level visibility.

This is where the conversation connects back to the broader Peak Property Performance® idea. In the Peak Property Performance® book, Bill Douglas and Drew Hall frame AI as useful only when the owner has a clear enough operating foundation to apply it. The tool is not the strategy. The strategy is knowing which data matters, where work breaks down, how decisions are governed, and how a better workflow improves the asset.

Start small, measure the result, and protect the owner from surprise complexity

One of Greg’s strongest recommendations was to start small and start practical. That does not mean thinking small. It means picking a workflow where the team can observe the pain, define the outcome, and measure whether the technology produced a meaningful improvement. In AP, that could mean fewer manual invoice touches, faster coding, cleaner routing, reduced approval delays, or better portfolio-level visibility into recurring expense patterns.

This matters because AI tools are not free to run, and poorly scoped implementations can create cost without operational gain. A chatbot, a document tool, or a general-purpose AI assistant may check a box internally, but if it does not solve a real bottleneck, the owner has simply added another vendor, another workflow, and another cost center. That is a familiar CRE problem. Every new tool promises clarity, but without a clear operating purpose, the portfolio gets another silo.

“Start small, start practical, start with something you can actually measure the result.”

Greg also made an important distinction between general-purpose tools and purpose-built products. General-purpose AI can be useful for broad tasks, but it may become expensive or brittle when forced into a specialized workflow. A CRE-specific AP platform, for example, is designed around the recurring patterns of real estate accounting. That does not make it automatically better for every owner, but it gives the evaluation a clearer standard: does the tool understand the job the team is trying to complete?

Owners can apply the same test across the property stack. If the issue is utility optimization, start with the building systems, meters, controls, and operating patterns that drive consumption. If the issue is insurance documentation, start with the data that proves risk management and maintenance discipline. If the issue is occupancy, start with the experience and services tenants actually feel. AI becomes useful when it is attached to a measurable operating play, not when it is purchased as a label.

Actionable takeaways for CRE owners evaluating AI

The pressure to “do AI” is real. Greg compared the current moment to earlier technology waves, when every company felt pressure to move online, adopt cloud software, or modernize customer systems. That pressure is understandable. No owner wants to fall behind. No executive wants to be the person who ignored a major market shift. But the lesson from this episode is that urgency should sharpen discipline, not replace it.

For CRE owners and asset managers, the first takeaway is to name the bottleneck before naming the tool. Where is the team spending too much time? Where are errors recurring? Where are approvals slowing down operations? Where does the property team lack visibility? Where does portfolio reporting require too much manual cleanup? Those questions turn AI from a broad mandate into a practical business conversation.

The second takeaway is to evaluate data readiness. AI-backed workflows need data they can consume, interpret, and act on under clear rules. That is true in AP, and it is true across building operations. If data is trapped in vendor platforms, inconsistent across properties, or disconnected from the systems that run the building, the owner’s ability to automate intelligently is limited. If you do not own your data & digital infrastructure, your vendors do.

The third takeaway is to connect each technology decision to owner control. OpticWise looks at this through the PPP 5C™ plan: Clarify, Connect, Collect, Coordinate, Control. Clarify the workflow and ownership gaps. Connect the systems and data sources that matter. Collect usable data in a consistent model. Coordinate access, privacy, lineage, and rules of use. Control which decision engines, vendor platforms, or internal tools can act under owner permissions. That is how Property Brain™ becomes Portfolio Brain™ over time.

The final takeaway is simple: do not let AI distract from operational fundamentals. Better technology should help the owner improve NOI, reduce risk, strengthen tenant experience, and make the portfolio more governable. In accounts payable, that means fewer low-value touches and faster, cleaner financial workflows. In building operations, it means IT + OT under an Owner Data Standard, so the owner can see what is happening and act with confidence. You can hear more conversations like this on the Peak Property Performance® Podcast.

About OpticWise: OpticWise provides owner-controlled data & digital infrastructure for commercial real estate — from PPP Audits to portfolio-wide intelligence. See how we operate or read customer outcomes.