Council Post: The Missing Link Between AI Investment And Revenue Growth
Eglae Recchia, CEO of Keyway, an agentic infrastructure built for CRE.
getty
For the last two years, model breakthroughs have dominated the AI conversation: bigger context windows, smarter co-pilots, faster reasoning and more autonomous agents. Inside enterprises, however, a different conversation is starting to take over that's far less theoretical and more operational: Where's the ROI actually coming from?
This has quietly become one of the most important debates in AI today. Companies are spending aggressively on AI while struggling to connect those investments to measurable business outcomes. Token consumption costs are exploding, AI pilots often break down inside real workflows, and teams are discovering that adding a generic assistant on top of fragmented operations doesn't automatically create operational leverage while also resulting in much less sustainable revenue growth.
Most executives I speak with are no longer asking whether they should use AI. They're asking why the economics still don't fully work. In industries such as commercial real estate, that gap becomes impossible to ignore because it was never a "chatbot problem" to begin with. It's an operational infrastructure problem.
Commercial real estate workflows are fragmented, document-heavy, collaborative and deeply dependent on institutional knowledge. Underwriting, diligence, lending, transactions and asset management all involve multiple stakeholders, disconnected systems and high-conviction decisions made under time pressure. In that context, the question is no longer whether AI can generate insights but whether organizations can actually operationalize those insights.
That distinction matters more than most people realize. Today, many companies can produce AI-generated summaries, automate portions of analysis or extract information from documents. However, the value often stops there. Teams still spend time reconciling data across systems, validating outputs manually, coordinating across silos and reworking processes that AI was supposed to simplify.
The result is an uncomfortable reality many executives are beginning to face: AI adoption doesn't necessarily translate into operational leverage.
This is where I believe vertical AI companies have an advantage. They don't necessarily have larger models or better marketing, and horizontal AI platforms don't lack technical sophistication. However, vertical AI companies generally understand the operational realities of the industries they serve.
For example, my company, Keyway, provides agentic infrastructure for commercial real estate. We understood that success with AI doesn't come from layering generic assistants onto existing workflows. Customized infrastructure that fits naturally into how real estate organizations actually operate means systems are designed around underwriting logic, diligence processes, lending workflows and real production environments.
The firms seeing the strongest results today are the ones redesigning workflows around connected systems where data, decisions and execution work together. This is the shift from AI as a feature to AI as infrastructure, and infrastructure changes the economics entirely.
When workflows become structured and connected, organizations begin to reduce operational variability. Underwriting becomes more consistent, decisions happen faster, risk surfaces earlier, and teams spend less time managing fragmented processes and more time executing. This creates scalability, which is much more valuable than productivity gains alone.
In today's environment, tighter capital conditions and increased scrutiny are forcing companies to operate with greater precision. In many industries, inefficiency is a real strategic disadvantage. In commercial real estate, for instance, the companies that can identify opportunities earlier, evaluate deals more consistently and move with greater confidence will increasingly outperform those operating on disconnected infrastructure.
In other words, the ROI comes from enabling organizations to do more business, faster and more intelligently. That challenge is the infrastructure behind the model and whether it can work inside enterprise production systems.
For years, the competitive advantage was having access to advanced models. Today, access is becoming increasingly commoditized. The differentiator is shifting toward operational integration, workflow intelligence and industry-specific infrastructure.
The companies that win will be those that can build systems capable of turning intelligence into execution reliably, consistently and at scale. Ultimately, AI alone doesn't create enterprise value. Operational infrastructure does.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?