Capitec Pulse works because Capitec invested in data architecture, data governance, and data quality long before the AI use case existed. What this means for organisations evaluating their own data foundations.
Capitec launched Pulse this week — an AI-powered contact centre tool that assembles a real-time picture of a client’s recent activity before the agent picks up the call. TechCentral and BizCommunity covered the launch, with the numbers: 18% reduction in call handling time, 26% net performance improvement across the pilot group.
But the numbers are not the lesson.
The lesson is what had to be true about Capitec’s data foundations for Pulse to work at all.
When Capitec’s CIO described the engineering behind Pulse, the most striking detail was not the AI model or the Amazon Connect integration. It was the years of foundational work that preceded it.
Capitec Pulse queries live operational data across the bank’s entire estate — over 60 databases — and assembles context in seconds. That is not an AI achievement. It is a data architecture achievement. The AI layer sits on top of foundations that were built long before anyone planned a contact centre use case.
Those foundations include:
Without these, Pulse does not work. Not because the AI fails, but because the data it depends on would be stale, inconsistent, or unreliable.
The immediate reaction from many executives will be: “We need something like Capitec Pulse.”
The honest assessment is that most organisations are not ready — and the gap is not budget or talent. It is data foundations.
Capitec’s CIO made this point directly: two conventional approaches to building Pulse — replicating data into a dedicated store, or querying production databases directly — both fail at scale. The first introduces stale data. The second degrades transactional performance. Capitec solved this because it controls its own source code and built its data architecture with these constraints in mind from the beginning.
Most organisations do not have that luxury. They run licensed platforms with limited architectural flexibility. Their data is spread across systems that were not designed to interoperate in real time. Master data is duplicated, inconsistently defined, or ungoverned.
The constraint is not technology. It is readiness.
What Capitec built — and what any organisation pursuing real-time operational intelligence must assess — are three interlocking capabilities:
Capitec standardised its entire estate on Aurora PostgreSQL with dedicated read replicas. Every database in the bank shares the same architectural pattern. This consistency is what allows Pulse to query across 60+ systems in seconds.
Most organisations have heterogeneous estates: different databases, different schemas, different replication patterns. Before any AI or automation initiative, the architectural landscape must be mapped and assessed.
Pulse works because Capitec knows which data lives where, who owns it, and how it flows. When the system assembles a client briefing, it draws on data whose lineage is understood and whose quality is maintained.
In organisations without governance, the same query would return conflicting results from different systems — with no way to determine which is authoritative. This is the governance gap that blocks automation, not just AI.
Real-time systems amplify data quality problems. A 2% error rate in a batch report is manageable. In a real-time agent briefing, it means one in fifty clients receives incorrect context — creating worse outcomes than no automation at all.
Data quality must be monitored, measured, and maintained as a continuous discipline. It cannot be assumed or retrofitted after deployment.
Capitec Pulse is a compelling example of what becomes possible when data foundations are in place. But the leadership question is not “How do we build Pulse?” It is:
Are our data foundations ready for the next opportunity — whatever it turns out to be?
When a new capability emerges — whether AI-powered customer intelligence, automated fraud detection, or real-time operational decisioning — the organisations that can act on it are those that already have:
The organisations that cannot act are those that discover, at the moment of opportunity, that their data is ungoverned, fragmented, or unreliable. The remediation timeline is measured in years, not weeks.
Most organisations know their data foundations have gaps. The challenge is understanding which gaps matter, what must change, and in what order.
This is where independent data advisory adds value — before platforms are selected, before vendors are engaged, before AI initiatives are scoped:
Capitec Pulse did not start with AI. It started with governance, architecture, and quality. The AI was the last layer — the visible output of invisible foundations.
The next technology shift will reward organisations that invested in those foundations before the use case arrived. Get in touch if your leadership team needs clarity on where your data foundations stand — and what must change before the next opportunity emerges.