Consumers are bypassing traditional search when it comes to finding holiday accommodation or properties to buy, preferring to use artificial intelligence (AI).
AI cuts out much of the manual work of clicking and reading reviews, instead returning a “humanised” answer that weighs up the options and makes a recommendation.
In response, companies are implementing agentic tools in their offerings that allow people to browse and select properties, estate agents to gather cross-cutting data to promote properties for sale, and a European bank is even enabling purchases through AI.
However, enabling AI to become a personal shopper comes with risks that, should the technology fall into the wrong hands, it can spend other people’s money and run up bills, says Mark Walker, director at technology consultancy T4i and former IDC South Africa MD.
According to property company Nox, Google Analytics for Cape Town shows that traffic to their rentals website is up 300% from ChatGPT, 100% from Gemini and Claude, and 90% from Perplexity.
At the same time, Naspers-owned Property24 ChatGPT, launched mid-year, lets users discover property listings across South Africa through natural conversation.
Only the relevant
Nick Taylor, MD of Nox Cape Town, says AI has changed how the agency structures its property data, descriptions and reviews. “AI assistants don’t respond to superlatives and flaky marketing.”
Instead, AI assistants “respond to specific, structured, verifiable facts. ‘Stunning villa’ means nothing to an AI, but a ‘six-sleeper with a heated pool, sea-facing terrace and dedicated workspace, 200m from Camps Bay beach’ is a recommendation waiting to happen,” says Taylor.
Almost half of bookings go to the first host to respond, with many guests abandoning their search within minutes if they don’t receive a reply, says Nox. AI can help speed up response time, but Taylor says a human who knows the neighbourhood is still vital to the booking experience.
“AI can’t tell you how the light hits Lion’s Head from a particular terrace at sunset, or that a ‘sea-facing’ balcony is glorious in March but battered by the southeasterly winds in December,” Taylor says.
Nox notes that more than 60% of operators globally already use AI, enabling real-time visibility of bookings, revenue and pacing, with plain-language summaries replacing raw spreadsheets.
Browse, compare, buy
Alistair Pugin, founder of @Celyntis, says people’s behaviour “is shifting from searching with Google or your booking app of choice, to using AI to find or recommend items because people are developing trust in AI to be accurate, better, faster and more comprehensive than a result returned from a search engine”.
Pugin explains that AI agents research, compare, negotiate and buy on a person’s behalf, while traditional search requires step-by-step human input and clicks, such as search, compare and reading reviews.
Walker says this shift in human behaviour makes sense as it allows for comparisons on price, service quality, location characteristics, reviews and weather in a single instance, rather than trolling through multiple individual sites that were pre-ranked by search-engine optimisation.
Companies such as Etsy and Shopify already use GenAI such as ChatGPT and Perplexity’s one-click shopping flow to enable check out, says Pugin. “The model is already moving beyond theory.”
The value
Lightstone says AI is reshaping property valuations in South Africa by combining trusted data, machine learning and human expertise, moving beyond valuation as professional judgement and historical comparables.
Its AI-powered tool, AiVM, uses features such as property location, size, previous sale price, and other structured inputs to estimate market value, says Hayley Ivins-Downes, managing executive of real estate at Lightstone.
Ivins-Downes says Lightstone’s data includes nationwide coverage of registered South African properties, millions of indexed properties and street addresses, and decades of deeds transaction history.
“By combining deeds records, cadastral data, municipal rolls, listing data, point-of-interest information and imagery, Lightstone’s system builds a much richer view of property value than traditional methods alone,” Ivins-Downes notes.
Pass go
Internationally, Realtor.com, owned by Nasdaq-listed News Corp, launched RealAssist AI in June. The AI home search tool built with Google’s Gemini and Google Cloud lets buyers search using natural language. Currently open to a group of beta users, it takes people through the process of finding a home, from initial search through to connecting with an agent and closing the deal.
Toronto-based TD Bank Group has launched an agentic AI model to automate mortgage and home equity line of credit applications, overseen by the bank’s Trustworthy AI team, which evaluates models on privacy, security, fairness, accountability and explainability before they reach clients.
Mastercard has piloted a solution with Spain-headquartered Banco Santander in Europe that lets an AI agent execute payments end-to-end, which it says is the first agentic payment carried out within a regulated banking framework. Santander ran the transaction in a controlled environment using Mastercard Agent Pay but did not disclose the nature of the transaction or purchase.
Matías Sánchez, Santander’s global head of cards and digital solutions, says the bank will move into extended testing and scaling; it has not since announced whether it has expanded its use of agentic payments. PayOS, a card-native payment and billing platform built for AI-driven transactions, handled the orchestration.
Share
