ARCE 2026 Speaker Spotlight: Antonia Dean on Why AI Should Be Treated Like the Intern, Not the Oracle

ARCE 2026 Speaker Spotlight

Accra, Ghana — 4 September 2026 · Property News Africa is an Official Digital Media Partner of ARCE 2026

Antonia Dean hears AI pitches for a living. As a venture investor focused on pre-seed and seed-stage technology companies, she has sat across from founder after founder pointing proudly to corporate pilots that never turned into contracts. The pattern bothered her enough to build a framework around it, and that framework — “AI as the Intern” — is what she will bring to the Africa Real Estate Conference & Expo in Accra on 14–16 October.

Her argument is simple and, for an industry that runs on relationships, paper and informal process, unusually practical: AI fails in most businesses not because the technology is weak but because it is dropped in like a tool rather than onboarded like a hire. Property News Africa put five questions to her ahead of the conference. Her answers are below, lightly edited.

Speaker snapshot

Where “AI as the Intern” came from

Property News Africa: Tell us briefly about your background and The Antonia Method, and how the “AI as the Intern” framework came about.

Antonia Dean: My path into venture wasn’t linear. I started at Harvard Law, then spent nearly a decade in cosmetics brand management, launching beauty products and brands for Calvin Klein, Michael Kors and Vera Wang, before moving into corporate VC at Comcast NBCUniversal. Today I invest in pre-seed and seed stage technology companies, and I’ve deployed more than $20 million into early-stage companies across my career to date. Alongside that, I run The Antonia Method, which brings together my keynote speaking, a weekly newsletter and an advisory practice focused on helping leaders make deliberate, high-agency decisions, especially around risk and technology.

“AI as the Intern” came directly out of what I was seeing from the investor’s seat. I hear pitches from AI companies constantly, and one pattern kept repeating: founders would proudly point to their corporate pilots, but those pilots weren’t converting into long-term commercial agreements. When I dug into why, the problem wasn’t the technology. It was how the technology was being brought in. Corporate leaders are often under pressure from a CEO or senior leadership to “just onboard AI,” so they choose a solution quickly, drop it into the organisation and expect results. What they’ve actually done is choose a tool, not onboard it. And then they’re disappointed with what they get back. That gap is where the framework was born.

What onboarding AI properly actually looks like

PNA: The framework’s core idea is that AI, like a brilliant graduate, needs onboarding rather than just workload. In practical terms, what does “onboarding AI properly” look like inside a business, and what does getting it wrong look like?

AD: Think about how you’d actually set up a brilliant new hire to succeed. You wouldn’t hand them your hardest, highest-stakes account on day one with no context and no supervision. You’d put them on the right team, give them assignments matched to their ability, set clear expectations for what good looks like, provide the context they need to make sound judgement calls, and build in guardrails and check-ins while they earn your trust. Onboarding AI properly is the same discipline. It means choosing the teams and workflows where the technology can genuinely help, being explicit about what success looks like, feeding it the context of how your business actually operates, and putting guardrails in place so that when it gets something wrong — and it will — the cost is contained and the lesson is captured.

Getting it wrong looks like the opposite of every one of those steps. It’s deploying the technology into the wrong teams, without expectations set, without the right guardrails, and without the context the tool needs to be useful. When that happens, the tool underperforms, people lose confidence, and leaders conclude that AI simply doesn’t work for their business. In almost every case I’ve seen, the technology wasn’t the failure. The onboarding was.

Where AI belongs in real estate — and where it doesn’t

PNA: Real estate is a relationship-driven, document-heavy and often informal industry. Where do you see AI adding real value first — development, sales and marketing, property management, valuation, investment analysis — and where should leaders be cautious?

AD: I’d reframe the question slightly, because for me it’s less about which part of the industry and more about which workflows matter. AI can drive real value in development, in sales and marketing, in property management, in valuation, in investment analysis — but only when it’s set against the right problem. So rather than asking “where in real estate does AI belong,” I’d ask leaders to look across their business for the workflows that are bottlenecked: the ones slowed down by manual processes, trapped in information silos, or spread across disparate, disconnected data. That’s where the value is, wherever it happens to sit.

Take relationships, since real estate runs on them. AI is not going to manage a client relationship for you, and it shouldn’t. But given the right context, it can surface which relationships need nurturing, which deals have gone quiet, which contacts you haven’t spoken to in months. It does the surfacing so you can do the relating. Where I’d be cautious is anywhere leaders expect the technology to exercise judgement it was never given the context to exercise, or to stand in for the human trust that actually closes deals. Use AI to remove friction and surface what’s hidden. Don’t use it to outsource the relationship itself.

Build, buy or wait?

PNA: How should a mid-sized African developer or brokerage, without a large tech team, decide whether to build its own AI capability, buy off-the-shelf tools, or wait?

AD: It comes down to a straightforward calculation: the cost of the option against the value it can realistically drive, weighed against your team’s internal capacity to support it. For most mid-sized developers and brokerages without a large tech team, building your own AI capability from scratch is the wrong first move. The cost and the capacity it demands are high, and the value stays uncertain until you actually understand how these tools behave inside your business.

I’d start with an off-the-shelf tool applied to a low-impact workflow. It’s the same judgement you’d use with an intern or an entry-level hire fresh out of university. You don’t hand them the deal that makes or breaks your quarter. You give them something real but low-risk, you watch how they perform, and you expand their responsibility as they earn it. Do exactly that with AI. Pick a bounded, lower-stakes workflow, learn what the technology does well and where it struggles, and build your own capacity and confidence from there. Waiting entirely carries its own risk, because your competitors are learning while you stand still. But learning through a small, well-chosen first assignment beats betting big before you understand the tool.

The one message for Accra

PNA: What’s the one message you want ARCE delegates to leave your session with, and who should make a point of attending?

AD: That onboarding AI doesn’t need to be scary. It’s a strategic business decision like any other you make as a leader, and it responds to the same discipline. When you define what success looks like and set the parameters up front, you shape the outcome. The organisations that struggle with AI are usually the ones that skipped that step. The ones that succeed treated it as a decision to be made well, not a threat to be managed.

As for who should attend: this session is for any leader who feels the pressure to “do something about AI” but isn’t sure how to do it well. Developers, brokers, property managers, investors — anyone responsible for deciding where AI fits in their business and accountable for the results. You don’t need to be technical to be in the room. You need to be the person who owns the outcome.

Editor

See Full Bio