Three actions to drive lasting ROI in the next phase of AI adoption

JLL’s 2026 Global Real Estate Investor Adoption AI Survey shows a gap between what investors want from AI and what they have achieved so far. Investors want AI to deliver topline revenue opportunities but are actually seeing the greatest benefits through operational efficiency. Investors should revisit their AI strategy and develop agentic workflows that create differentiation to their own advantage.

• Access to AI is now the baseline. 83% of investors have at least one AI activity scaled or fully embedded. 81% of all deployments use enterprise Large Language Models (LLMs).
• AI is delivering benefits, but not where investors set as top priority. There is a considerable mismatch between AI goals and results. Investors want AI to deliver topline revenue opportunities in investment but are seeing the greatest benefits in operational efficiency, particularly from asset and property management.
• Custom agentic AI workflows boost value. AI deployments built on custom agentic workflows are more than twice as likely to achieve four or more distinct benefits. Yet only 15% of investors have built agentic workflows in all use cases.
• Three actions to drive ROI: 1) Refocus AI strategy around ROI that lasts against competition, 2) develop the ability to build agentic workflows and 3) double down on your own advantages.
   

AI has spread through the commercial real estate industry at remarkable speed. Most investors now have access to it, and many can already see clear benefits: faster document processing, quicker analysis, easier report drafting. Far fewer can point to exactly how AI helps them achieve competitive advantage over their rivals. In a fast-moving market, improving from yesterday is not the same as pulling ahead.

JLL’s 2026 Global Real Estate Investor AI Survey covers 800+ investors across 12 countries, spanning institutional, private and listed investors and investment management firms. Of these, 80% are starting to see benefits from AI, yet 43% still struggle to achieve their goals.

Most notably, there’s a mismatch between what investors want from AI and what they are getting. Investors want AI to enable outperformance and to boost proprietary analytics capabilities, especially in capital deployment, the activity where they set the most AI goals. Where they actually see results is efficiency gains. And the strongest results so far come from asset and property management, where efficiency was the least prioritized goal.

This article sets out to dig into this phenomenon, examining the realities of AI adoption and the reasons behind this mismatch between vision and current results.

New agentic capabilities are shifting the AI conversation from individual productivity gains toward redesigning entire workflows and, in time, business models. This is the next phase of AI adoption, and most investors are entering it from the same starting point: fast, broad access to general-purpose tools.

Rollout is moving fast but it no longer signals maturity
 

The pace of rollout has been striking. In JLL’s 2025 Global Real Estate Technology Survey, 88% of respondents had started piloting AI or planned to start that year. By the end of 2026, 83% of investors have at least one activity area past the pilot stage, and 40% have every activity either scaling or fully embedded. This progress cuts across all seven activity areas.

However, successful AI rollout is no longer a reliable signal for true maturity because Enterprise LLM licenses, such as Claude and ChatGPT, provide a “shortcut”. Unlike earlier enterprise software, they don’t require as much integration, configuration or process change before going live. A company can roll out AI access widely without changing how the work is done, and the survey suggests this is exactly what’s happening.

While 81% of investors use LLMs for real estate, only 15% have built custom agentic workflows in all the areas they deploy AI. Agentic workflows are work processes in which AI systems carry out a sequence of tasks toward a defined goal, drawing on the firm’s own data and tools. One in five investors who have rolled out AI have not built any of these workflows in any of their target AI use cases.

This also leads to a less visible risk. When AI use stays at the level of individual experts, what they learn stays with them. It doesn’t become an organizational capability, and when those experts leave, much of that knowledge goes with them.

As use cases broaden, competitive pressures grow alongside them
 

Investors are using AI more widely than a year ago. The median investor now pursues seven AI use cases across five activity areas, up from five use cases in three areas in 2025. Capital deployment and investment management (92%) and data-related workflows (91%) are close to universal. Market trend analysis and investment thesis creation is the most common single use case, pursued by half of all investors.

With low levels of customization, convergence is underway. The most popular use cases draw heavily on market data that competitors can buy too, and they increasingly run on the same enterprise AI tools. When most firms apply similar models to similar data, the bar rises for everyone. Each firm gets better but it becomes harder to pull ahead.

Investors are already feeling it. 69% felt material pressure from AI’s impact. When they describe that pressure, competition is the most cited theme, alongside cost, talent and skills, and the pace of change. Almost all falls on the company’s own operating model and not market conditions.

Benefits are real, just not where investors want them most
 

Benefits from AI adoption start to become observable at the company level. These benefits span from more efficient operations to more accurate decisions, to cost reduction and revenue generation support.

Nonetheless, depth is still limited. On average, investors observe three of the six benefit types in each activity area. Only 4% achieve all six in any single activity and 15% see early signs of value but they aren’t yet measurable.

The larger issue is which AI use case areas these benefits are coming from. As shown earlier, there has been a considerable mismatch between what investors want to achieve and what they are currently achieving.

Investors’ top AI objectives are investment outperformance and proprietary analytics capability in capital deployment; each selected as a priority by 49% of respondents, yet these critical areas currently see the weakest results. Asset and property management shows the reverse pattern. This is seen as a lower priority area but has demonstrated the strongest results. It leads all other activities across efficiency (66%), decision quality (57%), capability (50%) and alpha (investment outperformance) (46%).

Why is this happening? Part of the answer is the goal itself. Alpha means outperforming competitors as well as the market, for example an index. It needs longer horizons to prove outperformance vis-à-vis benchmarks. It is inherently harder.

The earlier findings help explain the rest. When AI is rolled out mainly as general-purpose tools and used with the same purchased datasets and across similar use cases by all market players, its gains tend to be generic: Work gets faster and that shows up as efficiency improvements. Particularly with many companies capping token use to control costs, there is a hard limit on what can be achieved from individual use alone.

Achieving investment outperformance and capability building are more difficult to achieve. They depend on what a company can do which others cannot, and broad access to AI alone doesn’t provide it.

So, the most important question becomes: how can investors turn generic AI benefits into an advantage that lasts, especially in a competitive environment?

More of the same is unlikely to close the gap between what investors want from AI and what they get. A new approach is required to guide the next phase of AI investment. Three actions can sharpen the focus.

Action 1: Refocus your AI strategy around ROI that lasts, area by area
 

Most investors currently deploy AI across activity areas in the same way: To improve current operations, with investment split between quick wins and big bets. What’s often missing is a check on two things: Whether the activity involves external competition by nature and whether the return will last once competitors use the same tools. The answers to this determine what investors should focus on in that activity.

Where investors compete head-to-head, as bidders do when pursuing an acquisition, gains from AI can be competed away, especially efficiency gains. For example, if every bidder underwrites faster, no one stands out. Thus, differentiation in data and tooling should be the focus going forward.

Where investors assess themselves mainly against their own past performance, as in property management or asset management, there is no direct external competition. Gains from AI tend to stay with the company, where efficiency flows into net operating income (NOI) and asset value. Here, investors should focus on effectiveness and consistency. Making the tool work is the most important task. In these areas, investing further in differentiation adds little.

Investors should review their AI strategy through this lens, activity by activity, and ask of each AI investment whether there’s competition and whether its ROI would stand if every competitor used the same tool. Expectations and goals should realistically match each activity. For some areas, this means revisiting the goals and priorities themselves, not only the AI behind them.

Funding and resource allocation needs to be coordinated across activity areas as one holistic plan. This is particularly important in the current market cycle, where outperformance depends more on driving income growth than yield compression. Budgets are finite and every AI decision carries an opportunity cost.

Action 2: Develop the ability to build, and build smart
 

Most investors now access AI through enterprise LLMs, and only the minority build new, bespoke tools. The key dividing line between agentic workflow and enterprise LLM is enterprise-wide process versus individual tool use.

Building is where the extra effort pays off, and it doesn’t mean training models. An agentic workflow is the configuration and process integration layered onto purchased AI tools or general-purpose LLMs. Integrating AI deeper into the workflow is key to unlocking value creation.

According to the survey, deployments that include agentic workflows report every one of the six benefit types more often than those without, with the widest gaps in staff capability and capacity.

Building an agentic workflow forces a company to decide how the work should actually be done, which takes a set of capabilities beyond the tool itself:

• Redesigned workflows: the process rebuilt around the agent
• Ready, connected data: clean, governed data that agents can reach
• Context management: the right documents and rules for each task
• An orchestration platform: one layer to run and monitor agents
• Guardrails and governance: human checkpoints, permissions and audit trails
• Talent and ownership: named owners who build and supervise agents
   

Data readiness shows how much these foundations matter. That earlier 43% struggle rate is not evenly spread: Among investors whose data isn't AI-ready, 59% struggle to realize benefits, against 21% of those whose data is AI-ready.

This doesn’t mean an investor should now build everything in-house. A good mix of build and buy often delivers best performance across use case areas.

Making the build or buy decision is also a critical part of the capability package. JLL Tech Advisory (https://www.jll.com/en-us/guides/surviving-the-saaspocalypse) offers a working rule:
 

• Buy core systems of record such as IWMS, or specialty functionalities such as energy management systems
• Boost platforms where you need more than the vendor provides out of the box
• Build the proprietary AI workflows that create a durable edge
   

Most investors are taking actions to develop this capability. 88% have hired AI talent or a dedicated AI lead, 81% have contracted external AI advisors and 80% have done both. The task then is to decide how to use the talent and skills and what, exactly, is that durable edge to build for.

Action 3: Double down on your own competitive advantages
 

The key question here is whether your AI budgets really fund your own moat? It’s more common to fund visible, popular projects with direct deliverables, such as AI-supported market analysis for deal sourcing, even when an investor’s real differentiation sits elsewhere.

Investors need to be deliberate about the difference between using AI to make core strengths even stronger and hedging bets across lots of use cases and/or doing what everyone else is doing for fear of missing out.

The current problem is that few investors actually fund the AI that would strengthen these differentiations. Among investors naming “talent” as their moat, only 24% use AI to train and develop talent. Among those whose edge is “relationships and proprietary deal access”, only 22% fund AI for relationship management. A much higher percentage (43%) of the same group fund AI only in direct deal sourcing.

To find real outperformance, investors need AI that taps their own sources of advantage: investment track record, performance data, relationships and integrated systems across acquisitions, asset management and portfolio management teams. These use cases are sometimes less visible and pay off more slowly, but competitors cannot replicate them by buying the same tool. They’re the ones worth funding for investors aiming to outperform over the long term.

Closing the gap between what investors want from AI and what it delivers has become a question of strategy and capability more than procurement. Three questions can guide what comes next:

• Lasting gains: Which of our AI gains would remain even if every competitor used the same tool?
• Ability to build: Do we have it, where does building in-house give us an edge and where does “buying” remain the better strategy?
• Funding the moat: Is our AI budget funding our own differentiation, or just following the trend?
   

Access to AI has got the industry to where we are today. What comes next depends on how deeply investors build AI into their work, and so far only 15% have built agentic workflows in every area where they use it. As these workflows spread, that depth will matter more. What each investor builds with its own people, data and relationships will decide who turns AI into the outperformance they’re looking for.

The 2026 JLL Global Real Estate Investor AI Survey is a special edition of JLL’s Global Real Estate Technology Survey series. It examines how real estate investors are adopting AI, what results they are seeing and where those results fall short of their goals. It is intended for real estate investors, investment managers and the advisors who support their technology and strategy decisions.

This survey covers 837 senior decision makers (52% C-suite) at real estate investors with $1 billion or more AUM (43% over $25 billion, 8% over $100 billion). Investor types include institutional, private and listed investors and investment management firms. The survey samples companies across 12 countries in the Americas (191), EMEA (327) and Asia-Pacific (319).