AI in Commercial Property: Practical Uses Over the Next Three Years

What does AI in commercial property mean today?

The term covers a wide range of tools. Some are long established, such as software that learns patterns in energy data. Others are newer, such as language models that read and summarise documents or answer questions in plain English.

For property owners and managers, the useful way to think about it is as a set of assistants that handle tasks involving large volumes of information. Property generates vast amounts of paper and data: leases, contracts, invoices, inspection reports, utility readings, service requests and compliance records. People spend a great deal of time finding, reading and reconciling that material.

In my view, the sensible stance is balanced curiosity. AI is a capable tool for certain jobs and a poor fit for others. Predictions about the next three years are uncertain and the pace of change is fast. Treat any forecast, including mine, as informed opinion rather than fact.

Where will AI for building operations deliver first?

Several uses look realistic for Sydney buildings in the near term.

1. Lease and document analysis. Tools can extract key dates, rent review mechanisms, options, make good clauses and outgoings terms from large lease portfolios. This speeds up due diligence, audits and portfolio reviews. A lawyer or experienced property manager should verify the results, because errors in reading a clause can be costly.
2. Energy optimisation. Software can learn how a building responds to weather and occupancy, then adjust air conditioning schedules and set points to reduce energy use while maintaining comfort. These tools build on the building management system, which makes the quality of the underlying controls important.
3. Maintenance prediction. Analysis of equipment data can flag unusual behaviour in chillers, pumps and fans, prompting a repair before a failure. The benefit shows up as fewer emergency call outs and better planned capital works.
4. Tenant service. Chat assistants can answer routine questions, log requests and direct enquiries to the right person. They suit common queries such as after hours access or booking a loading dock, and they free staff for complex issues.
5. Reporting and insight. AI can draft summaries of performance data, compile sustainability reports and highlight anomalies in invoices or utility bills. Staff still need to check the output, though the time saved on first drafts can be considerable.
6. Space and occupancy analysis. Where sensors or booking data exist, AI can identify patterns in how floors are used and suggest changes to layouts or leasing strategies.

Each of these use cases has a measurable outcome, which is the test I would apply before investing. If a vendor struggles to explain how success will be measured, ask more questions.

Why does data readiness matter most?

Every AI tool depends on the quality of the information it receives. Tools applied to messy, incomplete or inconsistent data produce unreliable results, sometimes with convincing confidence.

Data readiness for property typically involves: * A digital document library. Leases, contracts and reports stored in an organised, searchable form. Scanned paper is a poor starting point. * Consistent building data. Standard naming for assets and data points, as discussed in my article on building technology integration. * A current asset register. Knowing what equipment exists, where it sits and when it was installed. * Reliable metering. Energy and water data at a useful level of detail, with correct labels and units. * Clear ownership. Someone responsible for the accuracy of each dataset. * Access rights. Agreement on who may use the data, including tenant information subject to privacy obligations.

A sound approach is to treat data readiness as the first project, rather than a prerequisite someone else handles. Cleaning and organising information delivers value on its own, and it prepares you for whatever tools emerge.

How should you govern AI and keep people in the loop?

Governance sounds dull, but it protects your organisation. Before using AI tools on portfolio information, agree a short set of principles.

Consider these questions: * Privacy. Does the tool process personal information about tenants, visitors or staff? Australian privacy obligations may apply. Take advice on your circumstances. * Confidentiality. Where does your data go when it enters the tool, and who can see it? Commercial leases and financial information deserve care. * Accuracy. How will outputs be checked? Language models can produce plausible but wrong answers. A reviewer should verify anything that influences a decision. * Accountability. A named person remains responsible for decisions, whatever a tool recommends. * Safety. Control systems that affect fire, life safety or critical plant need especially careful oversight, and AI recommendations should go through qualified engineers. * Vendor terms. Read how vendors use your data and whether they can train their models on it.

Human oversight is the principle that matters most. AI works well as a first draft, a second pair of eyes or a pattern spotter. Final decisions on legal, financial and safety matters belong with qualified people.

Staff also need support. Training, clear guidelines and an open conversation about how roles will change help adoption. People who understand a tool use it better and spot its mistakes.

How should Sydney owners start?

A practical sequence works well: 1. Choose one problem. Pick a task that is time consuming, repetitive and measurable, such as extracting lease data or reviewing utility invoices. 2. Check the data. Confirm that the information needed exists in a usable form. 3. Set a baseline. Record how long the task takes and how accurate it is today. 4. Run a small pilot. Test with a limited scope, a clear time frame and a named owner. 5. Review honestly. Compare results with the baseline, including errors caught, time saved and effort spent checking. 6. Decide next steps. Expand, adjust or stop, based on evidence.

Keep contracts flexible. The technology is moving quickly, and long commitments to a single tool carry risk. Prefer arrangements that let you export your data and change providers.

Finally, remember the sector's slower moving realities. Buildings last for decades, plant is replaced on long cycles and leases run for years. AI will change how information is handled sooner than it changes the physical building, which suggests that the biggest early gains sit in administration, analysis and operations rather than construction.

Final thoughts

AI in commercial property rewards practical thinking. Prepare your data, choose measurable tasks, set clear governance and keep qualified people responsible for decisions that matter.

If you would like more Sydney property insight, subscribe to my newsletter, The Fine Print, on LinkedIn and follow along.