JLL’s AI Strategy: Readying Systems for 2026

For a firm like JLL, leadership in the tech sector isn’t just about spotting prime real estate. It’s about building a digital infrastructure that doesn’t just crumble under the pressure of constant tech shifts. Getting your existing apps ready for AI isn’t a nice-to-have anymore. By 2026, it’s the baseline for staying competitive and efficient. So how do the top players actually get their old systems ready for what’s coming?

1. Conduct a Complete Application Inventory and AI Readiness Assessment

You can’t start aligning for AI until you have a dead-on accurate map of your current application field. And I mean a real deep dive into each system’s architecture, data dependencies, and integration points. I’ve seen projects go dead in the water for months because a team got sloppy here and ran into compatibility nightmares later. You have to document everything, from your main ERP systems down to the niche property management tools. A company like JLL, for instance, is juggling everything from big platforms like Yardi Voyager for property management to small, custom-built client portals.

For every single app, you need to check its AI readiness in a few key areas: data accessibility, API compatibility, and infrastructure flexibility. Can you actually get to the data? You need to know where it lives, its format, and how much work it’ll be to pull and prep it for AI models. Is it all in neat SQL databases or are you wrestling with a mess of unstructured docs and sensor feeds? Then, check the APIs. Do they even exist? Are they mature enough for an AI service to plug into? Lastly, look at your infrastructure. Can your current setup (whether it’s on-prem, cloud, or a mix) handle the heavy lifting AI requires? Something like ServiceNow’s Application Portfolio Management (APM) module is great for keeping this all in one place and helps you categorize apps by their technical state and business importance.

Pro Tip: Data Lineage Mapping is Non-Negotiable

Don’t skip data lineage mapping, a lot of people do, and they pay for it. You absolutely have to know the full journey of your data, from where it was born to where it sits now in every app. Mapping this out is how you spot hidden biases, quality problems, and compliance bombs before they poison your AI models. Whether you use specialized tools or just painstakingly detailed documentation, trace those data flows. If you don’t, your AI will be learning from junk, and you’ll get junk results.

2. Define AI Integration Use Cases and Prioritization Criteria

After you know what your apps can do, you need to figure out where AI will make a real difference. The goal is strategic application, not just cramming AI into every possible process because it’s trendy. For a company like JLL, this could mean completely changing the game in predictive maintenance for their buildings, figuring out how to use space better, or giving clients personalized property recommendations. It all starts by looking at specific pain points and asking, “could AI fix this?”

Pull people from all over the company, operations, finance, client services, IT, and get them in a room to brainstorm use cases. Your ops team might see an opportunity to use AI on sensor data to predict when equipment will fail, which slashes downtime and repair costs. The client services folks might want an AI chatbot to take care of the simple questions so their team can handle the hard stuff. After you have a long list, you need clear prioritization criteria to sort through it. I always use a matrix to score each idea on its potential business value (what’s the ROI?), technical feasibility (can we actually build it?), and data availability/quality. This gives you a ranked list. Knock out the quick wins first to build some momentum and prove this isn’t all just talk, then you can earn the right to go after the bigger, more complicated projects.

Common Mistake: Chasing Hype Over Value

I see this mistake all the time: companies jump on the AI bandwagon without a clear problem to solve or any idea of how it will actually help the business. What happens next is predictable. They burn through money on a pilot project that goes nowhere, and suddenly nobody in the company trusts any AI project ever again. Every single AI integration has to be tied to a real business goal with KPIs you can actually measure.

3. Establish a Strong Data Governance and Security Framework

Your AI is only as smart as the data you feed it. That means a solid data governance framework isn’t just a good idea, it’s non-negotiable. You need written-down policies for how data is collected, stored, and used, all aimed at ensuring quality and privacy. And with regulators breathing down everyone’s necks about data privacy, following the rules of General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA) is a legal necessity, not just good manners. If you’re a global company, this problem gets exponentially harder.

A good framework has to cover a few bases: data quality management to keep things accurate and consistent, and data security to control access and block threats. You also need to bake in data privacy (think anonymization and consent) and data ethics to make sure your AI is fair and accountable. Tools like Collibra Data Governance Center can give you a single source of truth for your data, which is exactly what you need to make sure your training data is clean and compliant. And for God’s sake, assign clear roles for data stewardship. If everyone is responsible, no one is.

4. Develop an Integration Strategy and Roadmap

Okay, so you’ve assessed your apps, picked your use cases, and set up your governance. Now you actually have to build the thing. Your integration strategy is the map that shows how the AI parts will talk to your old apps, what tech you’ll use, and in what order you’ll do it. My advice is always to go in phases. Start with smaller, simpler integrations so you can learn and fix things before you bet the farm on a massive project.

You’ve basically got two main ways to do this: API-led integration or data pipeline integration. The API-led approach is about using APIs to hook AI services (like NLP or computer vision) directly into your apps. A perfect example is an AI that scans lease agreements for key clauses by connecting to your document management system through an API. The other way, data pipeline integration, is all about pulling data out of your apps, cleaning it up, and then feeding it to your AI models. You’d use tools like AWS Glue or Google Cloud Dataflow to build out these pipelines. Whatever you do, your roadmap needs to have clear milestones, who’s doing the work, and when it’s supposed to be done.

Let’s make this concrete. If JLL wanted to use AI for predictive maintenance, their roadmap might be something like this. Phase 1: Get all the sensor data from their building management systems, like a Siemens Desigo CC, and pipe it into a single data lake. Phase 2: Use that data to train an ML model that can predict when an HVAC unit is about to die. Phase 3: Plug the model’s predictions back into their work order system so it automatically creates a maintenance ticket. Going step-by-step like this keeps the chaos to a minimum and lets you check your work as you go.

5. Implement and Monitor AI-Powered Solutions

Getting the solution deployed is just the start of the real work. After an AI solution goes live, it needs constant monitoring and tweaking. AI models aren’t static. They “drift” as the real-world data changes, and their performance will tank if you’re not actively managing them. You have to set up clear metrics to watch how your AI apps are doing. For that predictive maintenance system, you’d track the accuracy of its predictions or the drop in emergency downtime. For a chatbot, you’d be watching resolution rates and CSAT scores.

You’ll need tools for model monitoring and MLOps (Machine Learning Operations). Things like DataRobot MLOps or H2O.ai MLOps help you watch model health, spot weird behavior, and handle retraining. These platforms can automate a ton of work like data validation and versioning, which is what keeps your AI solutions working well over time. You also need to be doing regular audits of the AI’s decisions, especially for anything that has ethical weight, to make sure it’s being fair. The ongoing effort to maintain a good AI system is huge and often underestimated. It’s a constant cycle of deploy, monitor, evaluate, and do it all again.

6. Foster a Culture of AI Literacy and Continuous Learning

Adopting new tech is always about people, not just the tools. To make AI alignment work, you need a team that actually understands what AI can do, what it can’t, and the ethical tightrope you’re walking. That means investing in real training for everyone from the IT pros managing the servers to the business folks who will be using the AI-powered apps every day. And it’s more than just teaching technical skills. It’s about building a culture that is open to new ideas and actually uses data to make decisions.

A firm like JLL, for example, could run internal workshops like “AI for Property Managers” or “How to Use Predictive Analytics in Real Estate.” They could partner with schools or training companies to get their people certified in AI and data science. The real magic happens when you get cross-functional teams working together on these projects, which is the best way to break down the walls between IT and the business side. Not having enough internal expertise is one of the biggest things that will stop an AI program cold. Investing in your people is just as critical as investing in the tech, because in a sector that moves this fast, you’re either always learning or you’re falling behind.

Getting your apps ready for AI is a big job. It takes real strategic planning, disciplined data governance, and a culture that is serious about learning. But if you follow a structured, step-by-step process, a company like JLL can weave AI into its existing tech stack and find new ways to be more efficient and innovative.