At Retcon 2025, the buzz around artificial intelligence (AI) in commercial real estate (CRE) was impossible to miss. From tenant screening to portfolio optimization, AI tools are reshaping the industry—or so the hype goes. But in candid conversations with CRE leaders across asset classes, a different story emerged: frustration. While current AI solutions dazzle with potential, many fall short when faced with the messy reality of large, complex datasets. Hallucinations, inaccuracies, and an elusive “last mile” are leaving leaders dissatisfied—and searching for answers. Here’s what we heard, what it means, and why that tricky final 20% might just be the golden ticket for customized enterprise solutions.
The Complaint: AI’s Stumbling Blocks
Over coffee chats and panel Q&As, CRE executives didn’t hold back. They love the idea of AI—automation, insights, efficiency—but the execution? Not so much. Two recurring pain points stood out:
- Hallucination Headaches
“We fed our AI tool a dataset of lease agreements, and it spit out terms that didn’t exist—like a tenant with a 50-year lease at $1 a month,” one multifamily operator grumbled. Hallucination—when AI generates plausible but entirely fabricated outputs—isn’t just a quirky glitch; it’s a trust-killer. Leaders shared stories of AI confidently misinterpreting market trends or inventing property stats, forcing teams to double-check everything manually. - Inaccuracy with Big, Messy Data
“It’s great for small portfolios, but throw in 10,000 units or a mixed-use asset, and it’s a mess,” a portfolio manager sighed. Current AI tools struggle with the scale and complexity of enterprise-level CRE data—think sprawling spreadsheets, inconsistent lease formats, or submarket nuances. A retail leasing exec added, “Our AI predicted foot traffic trends based on outdated zoning data. It was useless for our next deal.” The bigger and knottier the dataset, the more these tools falter.
The consensus? Today’s off-the-shelf AI solutions—think chatbots, basic analytics platforms, or tenant screening apps—are hitting a wall. They shine with simple tasks but stumble when the stakes (and datasets) get higher.
The 20-80 Rule: So Close, Yet So Far
Here’s the kicker: these leaders weren’t dismissing AI outright. They saw the value—it’s just incomplete. As one industrial property owner put it, “The tech gets us 80% of the way there, but that last 20%? It’s the hardest part—and the most important.”
We heard this 20-80 split again and again:
- A multifamily VP said their AI rent verification tool nailed basic tenant checks but missed subtle red flags—like cash-for-keys settlements—that tanked renewals.
- An office leasing director praised their AI for targeting tenants but lamented its inability to model complex capital stacks for negotiations.
- A retail asset manager loved the predictive maintenance insights—until the system choked on integrating supply chain data for their warehouse portfolio.
The pattern was clear: AI handles the low-hanging fruit—repetitive tasks, surface-level insights—beautifully. But when it comes to the nuanced, high-stakes decisions that define CRE success, it’s leaving leaders stranded. That missing 20% isn’t just a gap; it’s where deals are won or lost, where portfolios thrive or stagnate.
Why the Last 20% Is So Tough
So why does AI stall out? Our conversations pointed to a few culprits:
- Data Complexity: CRE data isn’t clean or uniform. Leases vary by property, markets shift by submarket, and legacy systems spit out inconsistent formats. Off-the-shelf AI lacks the muscle to wrestle this chaos into coherence.
- Domain Knowledge Gaps: Generic AI doesn’t “get” CRE. It might crunch numbers but misses the context—like why a tenant’s payment history matters more than their credit score, or how zoning laws skew market forecasts.
- Hallucination Risks: Trained on broad datasets, these tools invent details when faced with ambiguity, a fatal flaw when precision is non-negotiable.
- Scalability Limits: Small-scale success doesn’t translate to enterprise-level needs, where thousands of variables and real-time updates overwhelm current models.
The result? A tool that’s 80% brilliant but 20% unreliable is, for many leaders, 100% frustrating.
The Opportunity: Customized Enterprise Solutions
Here’s where the light bulb went off: that tricky 20% isn’t a dead end—it’s a doorway. CRE leaders weren’t just venting; they were hinting at a need for something better. “If someone could tailor this tech to our data, our workflows, we’d be all in,” one exec said. Another nodded, “I don’t need a one-size-fits-all chatbot. I need a solution that knows my portfolio inside out.”
This is where customized enterprise solutions come in. Unlike generic AI, bespoke platforms can:
- Tame Complex Data: Built to handle your specific datasets—whether it’s 10,000 multifamily units or a mixed-use retail empire—custom AI cleans, structures, and analyzes with precision.
- Embed Domain Expertise: Trained on CRE-specific rules (think lease clauses, submarket quirks, or capital stack logic), these solutions avoid hallucinations by staying grounded in reality.
- Bridge the 20% Gap: From underwriting with 50+ workflows to supply chain forecasting for industrial hubs, custom tools tackle the high-value, high-complexity tasks off-the-shelf AI can’t touch.
- Scale Seamlessly: Designed for enterprise needs, they flex with your portfolio, integrating legacy systems and real-time feeds without breaking a sweat.
The vibe at Retcon? Leaders are ready to invest in solutions that go the distance—not just 80%, but 100%.
Implications for CRE—and a Call to Action
These conversations at Retcon 2025 aren’t just gripes—they’re a roadmap for the CRE industry:
- Trust Is Non-Negotiable: Hallucinations and inaccuracies erode confidence. AI providers must prioritize reliability, or risk losing the room.
- One Size Doesn’t Fit All: Generic tools work for starters, but enterprise CRE demands tailored precision. The future belongs to customization.
- The 20% Is the Money Maker: That final stretch—where decisions get complex and stakes skyrocket—is where competitive edges are forged. Ignore it, and you’re leaving value on the table.
- Opportunity Awaits: Companies that solve this—think xAI with its focus on robust, CRE-specific AI—could redefine the game.
For CRE leaders, the message is clear: don’t settle for “good enough.” Demand AI that nails the full 100%, not just the easy 80%. And for innovators? The plea from Retcon’s dissatisfied execs is your starting line. Build the tools that conquer complexity, banish hallucinations, and deliver results—because in CRE, close isn’t close enough.
