During his presentation at the 2026 IT Indaba, Nkgethu Tebogo Mampana, head of data management at Absa, unpacked the processes around building a trusted enterprise AI in a privacy-first world.
Using the metaphor of a human brain, Nkgethu explained how organisations should function using data as memory and cognition as ethical decision-making. “All of us have a brain and typically intent is a factor of using that,” he said.
“The difference with intelligence is whether that brain is fed the correct memory – data – and therefore the right attachments are made out of that.” This creates the ability to act through automated decisions that can be both trusted and explained.
Yet in reality, many companies are not perceived as trustworthy. “This is often due to disparate systems, conflicts that result from different versions of client data and a lack of shared definitions across departments,” Nkgethu said.
Companies therefore need to transition data from isolated technology tools to connected systems that create better customer experience, resulting in increased retention and reduced operational costs.
Connecting data to make meaning
“A connected enterprise brain gives us a choice about decision systems and not just another platform,” Nkgethu said. “Typically, when you have data in an organisation, it is done within different silos.” Yet these silos cannot function in isolation as this creates errors in risk assessment and decision-making.
“Effectively, we need data domains, groupings of data that allows us to connect these ‘neurons’ correctly to think better, faster and make the right decisions.” This helps us better determine what this data tells us enabling us to better learn from it.
When the data is there and no one brings it together, this creates conflict, creating different versions of the customer and risk. “This conflict is never about data, but rather missing context and clear meaning,” Nkgethu told attendees.
Ultimately trust is built when data is correctly understood and utilised, and forms part of a useful, decision-making system, essentially improving customer retention rates. This also leads to improved efficiency which results in faster decision-making and more productive employees.
This is why maintaining fairness, resilience and provenance in data can benefit company growth.
The human factor
"Automation is not accountability,” Nkgethu said. “There could be an AI war if we don't bind the way that we manage and regulate AI. So, we see that in 2027, regulators are starting to push for accountability to have inventive AI use and decision ownership," said Nkgethu.
Human judgement and active governance are therefore key in ensuring accountability. This is particularly important when decisions affect customers, involve significant risks, or produce unexpected results. Tebogo argued that organisations should therefore treat AI as a highly efficient employee, but one that requires strict supervision through monitoring evidence and client confidence.
Human involvement ensures that AI decisions are fair, explainable and aligned with customer expectations. Resilient organisations therefore require continuity in human judgement alongside advanced and adjustable technology infrastructure.
