Management skills property teams need to work with AI

I’ve handed over brand systems to associates who’ve never used a content management system (CMS), trained asset managers to review AI-drafted copy for rent documentation, and walked portfolio teams through their first agent-assisted investor reports. The pattern is consistent: the teams that get results from AI are the ones with existing management skills property directors already use to run projects, brief consultants, and review work. Direction, review, iteration, editing. The tools amplify what’s already there. If the management capability is thin, the output stays shallow. If it’s strong, the tools produce real work.

This isn’t about learning prompt engineering. It’s about applying the same management skills property teams use when briefing a junior planner, reviewing a photographer’s selects, or managing an external agency. Just faster, and at lower cost.

What management skills property teams need to work with AI

The core set is small: the ability to write a clear brief, recognise good work from adequate work, give actionable direction for a second draft, and know when to stop iterating. Those four capabilities determine whether an agent produces something you can use or something that sits in a folder.

Teams that already know how to set scope, review output, and give feedback move faster. Ethan Mollick, who researches AI adoption at Wharton, frames this as management becoming a superpower in organisations using agents. The ones that don’t have these capabilities generate volume without utility.

Why do property teams with weak management capability struggle with AI?

A pitch brochure due in three weeks doesn’t write itself. Neither does an AI draft it well if the brief is “write something compelling about our mixed-use scheme”. Weak direction produces weak output. No amount of tooling compensates for unclear scope, vague success criteria, or an inability to articulate what’s wrong with draft two.

I see this during handovers. If the associate can’t explain what makes a good hero image for the homepage, or what tone the leasing brochure should carry, the agent won’t guess it. The output will be generic, requiring more rework than if a human had written it from scratch.

What does inadequate briefing look like in practice?

The brief lacks specifics: no audience definition, no length constraint, no tone guidance, no examples of what good looks like. The reviewer can’t explain why draft one missed the mark beyond “it doesn’t feel right”. The feedback loop stalls because the direction isn’t actionable. That’s a management problem, not a technology problem.

How should property teams build AI literacy in 2025?

Start with tasks that already have clear quality thresholds. Summarising a planning document. Drafting an internal update. Reformatting rent documentation for a board pack. These are bounded problems where good enough is defined, and review takes minutes rather than hours.

Run a few cycles with an agent. Write the brief. Review the output. Give feedback. Iterate. Document what worked. That loop builds the management skills property teams need before moving to higher-stakes work like investor narratives, public consultation materials, or pitch documents for a stalled scheme.

BCG’s research on AI and productivity found that performance diverged sharply between users who treated the tool as a co-worker requiring direction and those who treated it as a search engine. The former group iterated, reviewed, and edited. The latter group copied and pasted.

MIT’s research on generative AI and skilled workers showed similar patterns: performance gains came from those who already had domain expertise and editorial judgement, not from those hoping the technology would substitute for capability.

What does training for property teams using AI actually cover?

Not prompt templates. Not tool demos. The structure is: brief an agent to draft copy for a specific piece of collateral, review the output against your quality standard, mark what’s wrong and why, re-brief for a second draft, and decide when it’s done. Then do it again with a different task.

I run this on-site with asset teams during handovers. The associate who’s inheriting the brand system drafts a leasing one-pager using the agent, applies the tone guide I’ve documented, checks it against the visual system, and decides if it’s ready or needs another pass. That’s the capability transfer. Not a login and a Loom video.

Why does handover need to be engineered for capability transfer from day one?

If the brand system only exists in my head, the next associate can’t run it. If the content rules aren’t documented, the agent has no guardrails. If the tone of voice is “you’ll know it when you see it”, review becomes subjective and slow.

I document file sources, fonts, image specs, copy structure, approval thresholds, and the edge cases that came up during the project. That documentation is what lets the in-house team manage agents after I’ve handed over. Without it, they’re guessing. With it, they’re directing.

This applies to photography selects, signage specs, hoarding content, and investor report structures. If it’s not written down with enough precision that someone else can reproduce the standard, it won’t transfer. And if it doesn’t transfer, the tools don’t help. These are fundamental management skills property teams rely on whether working with people or AI.

How should small property teams start adopting AI in 2025?

Pick one repeating task with a clear output standard. An asset update that goes to the same distribution list every month. A pre-meeting summary of leasing enquiries. A planning document that needs reformatting for the board. Something where you already know what good looks like, and where review takes ten minutes, not two hours.

Run it with an agent for three cycles. Document what worked in the brief and what didn’t. Adjust. If the output quality is consistently acceptable, keep it. If it’s not, either the task is too ambiguous or the review criteria need tightening. Either way, you’ve learned something about your own management skills property teams rely on to get work done.

The teams that start here, with bounded scope and existing standards, build confidence and capability. The ones that start with “let’s use AI for everything” generate noise and abandon the tools within a quarter.

What happens when property teams treat AI as a direct report?

The metaphor is useful. You wouldn’t give a graduate planner an instruction like “make the scheme sound good” and expect a usable draft. You’d brief the positioning, the audience, the length, the tone, and the three points that must land. Then you’d review, mark up, and send back for revisions.

That’s the same process for agents. Brief with specificity. Review against a standard. Give feedback that can be acted on. Iterate until it’s done. The management skills property directors use every week with consultants, contractors, and junior staff apply directly. The tools are faster and cheaper, but they’re not autonomous.

Anthropic’s research on constitutional AI highlights that models perform better when given explicit rules, constraints, and examples. That’s management. The agents don’t self-improve without direction. They get better when someone with judgement steers them.

Why isn’t AI literacy for asset managers a software question?

It’s a judgement question. Can you tell when a draft leasing narrative undersells the transport links? Can you spot when an investor update is too vague about the vacancy on the second floor? Can you articulate why the tone in a planning submission feels defensive rather than confident?

If the answer is yes, you can manage an agent producing that work. If the answer is no, the agent won’t fix it. AI arms teams, it doesn’t replace the editorial judgement that separates adequate work from work that moves a decision forward.

I see this during training and handovers. The asset managers who already have a point of view about their collateral, their positioning, and their audience adapt quickly. The ones who’ve outsourced all creative and editorial judgement to agencies struggle, because the agents need direction they’re not used to giving. This is where strong management skills property professionals have developed become critical.

What documentation outlives the engagement?

File sources. Brand guidelines that specify not just colour and type, but tone, image style, and hierarchy. CMS instructions that assume no prior knowledge. Content rhythms that explain what gets published when, and who approves it. Edge case decisions: what I did when the planning image wasn’t available, or when the leasing agent requested off-brand layouts.

This is what lets the next associate manage the system without me. And it’s the same documentation that lets them manage agents. Specificity transfers. “Make it look premium” doesn’t. “Use Founders Grotesk Medium, 24pt, sentence case, aligned left, with 40pt leading” does.

The property teams that get lasting value from AI are the ones that already document how things should be done. The ones that rely on institutional memory and informal review find the tools harder to adopt, because there’s no written standard for the agent to follow or the reviewer to apply.

Frequently asked questions about management skills and AI adoption

Do property teams need technical skills to work with AI?

No. The management skills property teams need are the same ones used to brief consultants and review work: clear direction, quality standards, actionable feedback, and knowing when output is done. Technical skills aren’t the constraint. Management capability is.

How long does it take for asset managers to become confident using AI agents?

Three to five cycles on a bounded task with clear quality thresholds. If you can already articulate what good looks like and give specific feedback, confidence builds quickly. If those skills are weak, the learning curve is steeper because you’re developing management capability, not just learning a tool.

Can AI replace the need for external agencies or consultants?

No. AI arms teams, it doesn’t replace them. Agents need direction, review, and editorial judgement. They amplify existing capability but don’t substitute for expertise. Teams with strong internal management skills property professionals use can bring more work in-house. Teams without those skills still need external support.

What’s the biggest mistake property teams make when adopting AI?

Starting with high-stakes work or vague scope. A pitch brochure for a major scheme isn’t the place to learn. Start with repeating tasks that have documented standards and low consequences if the first draft needs rework. Build capability on small problems before tackling large ones.

Should training focus on specific AI tools or general principles?

General principles. The management skills property teams need, writing clear briefs, reviewing against standards, giving actionable feedback, transfer across tools. If training is tool-specific, it becomes obsolete when the software changes. If it’s principle-based, it lasts.

If you’re preparing to hand over a brand system, position an asset for sale, or train an in-house team to run their own marketing rhythm, the management skills property teams develop during that process transfer directly to working with AI. I’ve written about that handover approach before, and the AI automation systems I build follow the same principle: capability transfer, not dependency. The brand work and digital projects I run all include documentation that outlives the engagement. If you’d like to talk through how training and capability transfer would work for your asset team, send a paragraph. I’ll write back.