Artificial intelligence is impossible to ignore in real estate right now but knowing how to approach it strategically is another matter entirely. At FlexSA’s Annual Conference, Yardi’s Mandy Jhamat sat down with Zoe Webster, AI and innovation consultant at Authentic Innovation, to discuss how to meaningfully embed AI in flex operations.

Start with strategy. Yours, not AI’s

The most common mistake organisations make, according to Webster, is letting the technology lead. “Often in many organisations it becomes, ‘we must do AI and here is what we’re going to do’,” she said. “But let’s stop. What is it that is really important to the organisation?”

Her framework is deliberate and sequential – begin with your corporate strategy, understand what you are genuinely trying to achieve over the next three to five years, and only then ask which elements of that ambition could be meaningfully enabled by AI. From there, a data strategy and an AI strategy should follow, not precede, that clarity of purpose.

She was quick to acknowledge this is tidier in theory than practice. “It is a little bit messy,” she admitted. “It is not one size fits all.” The three strategies – corporate, data and AI – are in constant conversation with each other, each informing and updating the others as technology and business conditions evolve.

“AI is nothing without data. But a data strategy alone won’t save you if you don’t know what business problem you’re trying to solve.”

Zoe Webster, Authentic Innovation

Data: the foundation most organisations are still building

The survey statistic that drew the most visible reaction in the room was the 86% who cited data fragmentation as their main AI blocker. Webster was unsurprised. Even organisations that know where their data lives often struggle with access, ownership, storage costs and quality. “Who owns that data, who has responsibility for it, whose budget is covering that storage? It gets quite involved quite quickly,” she said.

For organisations earlier in their data maturity journey, she recommended starting with an honest audit – what do you have, where is it, what quality is it and can you actually access it? First-party data you hold alongside relevant third-party data are both worth assessing. The findings will shape which AI ambitions are realistic now and which need to wait.

AI literacy is a people strategy, not a training programme

Webster is clear that AI adoption succeeds or fails on culture, not tooling. Her goal when working with organisations is not to produce uniform enthusiasm for AI, but to give people enough understanding that they can engage with it critically and contribute their own ideas from the ground up.

“Often you see a top-down approach – you will use this new AI tool and you will be more productive,” she said. “What really works, for the long term, is getting people in the organisation comfortable enough that they can come up with their own ideas and take those to leadership.”

What good AI literacy looks like in practice:

• People have a shared vocabulary and can discuss AI meaningfully across teams
• Staff understand at a high level why AI excels at certain tasks and falls short at others
• Both enthusiasts and sceptics are brought into the conversation and their combination surfaces both opportunity and risk
• People feel empowered to propose AI applications from the ground up, not just directed from the top
• Shadow AI use decreases because people feel clear on what is acceptable and why

On the question of whether everyone in an organisation should reach the same view of AI, Webster pushed back: “I think it’s healthy that people have different perspectives. The magic happens when you bring those people together.” She noted that the most sceptical voices often do the most valuable work, asking the questions about risk and bias that the enthusiasts overlook.

Governance: accountability needs a name attached to it

When it comes to governance, Webster’s advice was unambiguous – someone specific must own it. “Often you get, everyone is accountable – and in that case, no one is,” she said. “When something goes wrong, who is responsible? Unless someone’s name is down, nothing really happens.”

She recommended that organisations review existing IT, information security and acceptable use policies to assess whether they already address AI and to supplement them with an explicit AI policy that sets out both what is permitted and the principles guiding every decision. The UK government’s five AI principles offer a useful starting framework, particularly around accountability, transparency and fairness.

Fairness, she was careful to clarify, cuts two ways – it means testing AI systems for bias and discriminatory outputs, but it also means ensuring equitable access to AI tools and literacy across the whole organisation, not just for those already closest to the technology.

“In the absence of policy, people don’t stop. They just do it in the shadows. Either way, that’s a problem.”

Zoe Webster, Authentic Innovation

Build or buy? The honest answer

The question of whether to build bespoke AI solutions or procure existing ones came up from the floor, and Webster gave a characteristically practical answer. Building your own large language model or end-to-end AI management system rarely makes sense for most organisations, but it might, under specific conditions. For example, if your data is too sensitive to share with any vendor, if you see a genuine path to commercialising what you build, or if your need is genuinely too bespoke for anything available off the shelf.

The part most organisations underestimate, she cautioned, is what comes after the prototype. “Building it is almost the easy bit. The harder part is the ongoing operations and maintenance and the opportunity cost of the time and energy you have taken away from your core business.” Anyone considering a build route needs to be honest about whether they can sustain it.

Sustainability: a risk hiding in plain sight

When the conversation opened up for Q&A, an audience member raised the question of sustainability. Webster acknowledged directly that AI is computationally intensive and energy hungry, and that at present, the transparency needed for organisations to accurately measure the carbon cost of their AI use simply does not exist. “There is a real risk that what you’re doing with AI runs counter to any kind of corporate sustainability objectives.”. She advised organisations to stay aware of the tension and be open with their people about it and expressed hope that regulation will eventually force greater disclosure from providers.

One practical piece of advice

When asked to close with a single actionable recommendation, Webster offered something elegantly simple – test your AI strategy on the person in your organisation who is most sceptical or most uncertain. “Does it resonate with them? Do they understand their path? Or do they shrug and say, what does this mean to me?” If the strategy only makes sense to those who are already bought in, it is not yet ready to guide the organisation as a whole.

As a secondary signal to watch for, whether people are openly sharing what works and what does not. “Find someone who’s maybe not sharing or is asking lots of questions and test the strategy on them.” In Webster’s view, a culture of honest knowledge-sharing about AI, including the failures, is one of the clearest signs an organisation is on the right track.

Explore more and speak to a member of our team about how Yardi Kube can support flex operators with the tools and data infrastructure to make AI adoption a strategic reality.