For many South Africans, a home is more than an address. It is where families build security and plan for the future. So when artificial intelligence starts influencing property valuations, bond decisions and financial risk, the question is bigger than whether the technology is clever. Can people trust the decisions it makes about their lives?
That was the central conversation at Lightstone’s recent 20 Years in Banking – The Trust Algorithm conference in Johannesburg, where industry leaders considered how data and AI could reshape South Africa’s property and financial services sectors.
The real question is who can challenge the algorithm
AI is moving beyond analysing information and making recommendations. Increasingly, systems can initiate processes and take action with limited human involvement. That may improve efficiency, but it raises urgent questions: Who approved the decision? Can people understand why it happened? What if the information is wrong? Can the outcome be appealed or reversed?
Technology analyst Arthur Goldstuck said that as AI enters its “agentic” phase, automated actions must be traceable, explainable and reversible. In property and finance, where decisions can affect access to credit and the security of a family home, accountability cannot be optional.
More information alone will not solve this. Lightstone co-founder and Simply CEO Anthony Miller argued that more data does not automatically create more trust. The value lies in turning it into accurate intelligence that can be checked and understood.
Bringing the invisible property market into view
The bigger opportunity may be connecting a property ecosystem where municipalities, banks, insurers, estate agents, valuers and consumers hold separate pieces of the picture.
Lightstone says around 60% of South African households sit outside the Deeds Registry and remain largely invisible to traditional property market systems. When homes and communities are missing from formal datasets, residents can struggle to have their assets recognised by systems used to assess property value, risk and access to finance.
Aerial imagery, computer vision, municipal records and alternative credit data could help build a fuller picture. Used responsibly, these tools may create pathways towards formal funding for people whose properties have historically been difficult for institutions to assess.
But better visibility must translate into fairer treatment – not simply more surveillance or faster decisions made without people’s input.
Earlier support, not just faster rejection
Connected data could help lenders spot signs of financial pressure before a missed bond payment becomes the first warning. Changes in municipal rates payments or other financial patterns may give institutions a chance to reach out sooner and understand what is happening.
That could make property finance more responsive, provided the signals are reliable and customers are treated with context and care. RMB chief economist Isaah Mhlanga noted that traditional economic indicators can lag behind fast-moving realities.
Why human judgement still matters
More automation does not make human judgement less valuable. Nedbank’s Josephine Baliah highlighted the importance of human ingenuity, connection and shared values. People bring empathy and context that data alone cannot fully explain.
Sara Winstone, Managing Director of Lightstone Property, said trust is built through consistent behaviour, robust methods and a willingness to be challenged. That is a useful standard for any organisation bringing AI into decisions that affect the public.
The next era of property intelligence should not be measured by how many decisions machines can make. It should be measured by whether those decisions are understandable, accountable and improve people’s options. In a country where property ownership and access to finance remain uneven, the most powerful innovation may be technology that earns the confidence of the people it serves.







