Management skills property teams need for AI tools

An asset manager needs a pitch document in three weeks. The deck exists, the data exists, the photography exists. What doesn’t exist is the coordination layer between the person who knows what needs saying and the AI tools that can draft, format and produce it. The constraint isn’t capability. It’s the management skills property teams were never trained for, because until now, you either did the work yourself or briefed someone else to do it. AI sits in between, and most teams don’t know how to manage it.

Ethan Mollick, writing in One Useful Thing, argues that management becomes the differentiating skill in a world of capable agents. The people who know how to decompose tasks, assess quality and orchestrate workflows will outperform those who treat AI as a magic box. That insight applies directly to property teams trying to run leaner without losing control of brand, narrative or output quality.

What management skills do property teams need to run AI tools effectively?

Three things matter: task decomposition, quality assessment and workflow design. Task decomposition means breaking a pitch document or planning submission into twenty constituent parts, each of which can be assigned to a tool, a person or a combination. Quality assessment means knowing what good looks like, so you can tell when the output needs another pass. Workflow design means deciding what happens in what order, and where human judgment sits in the sequence.

Most property teams have none of this muscle memory. An asset manager might know the portfolio inside out, but has never had to brief an AI model on tone, structure and constraints for a draft that goes to the investment committee on Tuesday. A leasing manager might commission photography twice a year, but has never had to evaluate AI-generated copy for a hoarding going up on Friday. The skills required are adjacent to what they already do, but not identical.

Why can’t property teams just learn the tools as they go?

Because the tools change faster than institutional memory forms. A portfolio team that spent six months learning one platform will find it deprecated, merged or outpaced by the next time work comes round. What doesn’t change is the underlying skill: how to break work into parts, how to specify what you need, how to evaluate what comes back and how to integrate it into the workflow that existed before AI arrived.

Training the tool is a waste. Training the management layer around it is durable. That’s why my training and implementation services focus on task structure, quality frameworks and handover documentation that outlives the engagement. If the next associate can’t run the system without me, I haven’t done the job.

What does AI literacy look like for an asset manager?

It looks like knowing which tasks are compressible and which are not. Writing the first draft of a leasing brochure: compressible. Negotiating the lease: not. Formatting twenty property profiles into a consistent template: compressible. Deciding which properties to include: not. An asset manager with AI literacy can look at their own workload and draw that line, then route the compressible tasks to the right tools and keep the judgment tasks in-house.

Research from Boston Consulting Group and Harvard Business School found that consultants using AI without training performed worse than those using no AI at all, because they delegated the wrong tasks and didn’t check the output properly. The same risk applies to property teams. Give someone access to a capable model without teaching them task decomposition or quality assessment, and they’ll either avoid it entirely or trust it too much. Both are expensive.

How should a small property team start adopting AI in 2026?

Start with one repeating task that’s time-consuming, formulaic and low-risk if it goes wrong. Formatting investment committee reports. Writing first-pass property descriptions. Preparing planning consultation documents. Pick one, document the current process, then redesign it with AI in the middle and a human review at the end. Run it three times, refine the workflow, then train the next person to run it without you.

The mistake most teams make is trying to adopt AI everywhere at once, or waiting for a perfect use case that never arrives. The better approach is to build competence in one workflow, then replicate the structure across others. That’s why I run on-site training sessions with asset teams, walking through their actual workload and identifying the five tasks where AI automation saves time this quarter, not in some theoretical future state.

What should the handover documentation include?

Everything the next associate needs to run the system without calling me. File sources, font licences, CMS login credentials, content rules, image specifications, approval workflows and the task-level instructions used to generate each piece of collateral. If the leasing manager leaves and someone else inherits the portfolio, they should be able to open the documentation and produce the next brochure to the same standard.

Most agencies don’t document this way because they’re not incentivised to make themselves redundant. I am. My engagements are time-limited and designed for the client to own and operate the system after I leave. That includes the AI workflows. If you can’t run it without me, the capability hasn’t transferred and the value hasn’t compounded.

Who on the property team should own this?

Whoever owns the output. If it’s a pitch document, the asset manager owns it. If it’s leasing collateral, the leasing manager owns it. If it’s the fund report, the fund manager owns it. The mistake is delegating AI adoption to IT or operations, because they don’t have the domain knowledge to assess quality or redesign the workflow. The person who signs off the work is the person who should learn to manage the tools that produce it.

That doesn’t mean they have to become prompt engineers or learn Python. It means they need to understand task structure, model capability and quality control well enough to integrate AI into their existing responsibilities. According to MIT Sloan Management Review, most of that is teachable in a day. The rest is practice.

If your team is trying to work out where AI fits, or you’ve adopted tools but nobody’s confident running them, I train for that. Send a paragraph about what you’re working on, and I’ll write back.