AI in Portfolio Management: What Are We Actually Using It For? And What Could We Be?

The business case for AI in commercial real estate is no longer theoretical, but the gap between promise and scaled deployment remains wide. The Ops, Data & Tech Tribe at CREAM UK 2026 cut through the hype to ask what the industry is actually using AI for, and what it could be. The answer: the potential is real, but it sits behind unglamorous prerequisites - chief among them clean data - alongside unified platforms, proper training, and a firm "human in the loop," with the most tangible wins coming from targeted, task-specific tools.

Tribe Host: Matt Smith (Yardi)

Co-Hosts: Ruman Sahota (Legal & General), Jonathan Theobald (Derwent London), Jenny Obee (Related Argent), Matt Warren (Lendlease), Jacinta Rowsell (Canary Wharf Group), Jessica Berney (The Workspace Group) and Charis Wells (Freedom Works)

Summarised by: Edgar To

Clean Data as the Essential Foundation

Clean data is the essential foundation, and the point on which most of the discussion turned. Before advanced AI agents can be deployed successfully, organisations must fix their underlying data infrastructure. Integrating fragmented data from finance, sales and building management systems into a single platform frequently exposes serious formatting and accuracy problems - and the principle of "bad data in, bad data out" makes data cleansing a significant and time-consuming first step that cannot be skipped.

From Silos to Seamless Integration

That foundation enables a wider move from silos to seamless integration. There is a strong push to break down legacy data silos using open APIs to connect operational software, so that systems can talk to one another. Linking building access data such as door swipes with the sales CRM, for example, allows organisations to trigger proactive alerts - identifying a tenant attending the office less frequently who might be a flight risk.

Governance, Training and the Human in the Loop

Governance and training featured heavily, organised around the idea of keeping a human in the loop. The Tribe stressed the need for formal AI training, particularly in effective prompting, and raised growing concern about staff relying blindly on AI outputs or dropping AI-generated text into important documents without checking it. The prevailing approach is to treat AI as an assistant rather than a replacement, supported by clear policies that ensure professional oversight remains.

Task-Specific Tools Over General Use

In practice, success is coming from task-specific AI rather than general use. A striking example was the use of AI-assisted drones for building façade inspections, which compressed a weeks-long, high-risk process into a single day and significantly reduced service charge costs.

The Next Generation and the Skills Gap

A substantial debate emerged over the skills gap: some participants warned that automating foundational "grunt work" - reading complex leases, running basic models - could erode the expert intuition and judgement that junior staff have traditionally built over time, while others argued that newer entrants adapt to AI seamlessly, delivering far higher output and spending more time on judgement and strategy than on manual data entry.

Biggest Opportunity: Proactive Operations and Efficiency Gains

The strongest opportunity lies in driving proactive operations and substantial efficiency gains. By connecting clean, integrated data with targeted AI tools, organisations can shift from reactive management to genuinely proactive operations - predicting customer retention risk, equipping sales teams with immediate portfolio insight, or enabling a junior employee to complete days of underwriting analysis in a single hour. The combined effect is a marked reduction in operational cost and the freeing of human capital for higher-value, strategic decision-making. The prize is not AI for its own sake, but a more responsive and efficient operating model built on data the business can trust.

Biggest Challenge: Avoiding "Pilot Purgatory" Amid Constant Change

The principal challenge is navigating the hyper-fast pace of change without falling into "pilot purgatory." The technology is evolving so quickly - sometimes week to week - that defining firm, long-term success criteria is genuinely difficult. Organisations are wary of backing the wrong provider or becoming trapped in an endless cycle of testing, made worse by nervousness around integrating historical data, meeting regulatory requirements, and accommodating shifting capabilities. The risk is that caution hardens into paralysis, leaving promising tools stuck in perpetual trial rather than scaled into the business.

CREAM UK took place on 23 June 2026 at Fulham Pier, London.

Find out more about the event at space-plus.org/cream-uk