Skip to main content

About

Domain expertise is not a credential.
It is thirty years of accumulated instinct.

Most AI engineers building for South African financial services have read the compliance manuals. Thirty years inside the industries being automated. Not studying them. Inside them — on the phones, in the branches, in front of regulators, running the P&L, signing the compliance reports. That is what domain expertise means here.

That expertise is not stated in the proposal. It is baked into every data model, every compliance guardrail, every workflow decision, and every edge case that gets handled before a client discovers it.

Centurion, Gauteng, South AfricaCenturionAI · Founded 2024

What domain expertise actually is

Five layers. All of them built over decades — not months.

Domain expertise is not what you studied. It is not your certifications. It is the accumulated ability to walk into a problem in your field and immediately know things that a competent outsider would take months to learn — if they learned them at all.

It is what separates someone who can talk about an industry from someone who can actually navigate it under pressure. A person with genuine domain expertise can spot the problem a layperson would completely miss, propose solutions that are realistic given the actual constraints of that specific industry, and avoid costly mistakes that come from not knowing the unwritten rules.

Foundational knowledge

The core principles, regulations, and terminology. FAIS, POPIA, BCEA, NCA, the Administration of Estates Act — not looked up when they become relevant. Known and applied from years inside them.

Applied experience

Having actually worked in the field. Made decisions. Seen what breaks in the real world under real pressure. Knowing what advisors do when they are under target. What regulators actually push back on versus what the manual says.

Pattern recognition

Reading a situation quickly because you have seen it before. A compliance risk that looks routine. A workflow that will break at scale. An assumption buried in a client brief that will cost money later.

Tacit knowledge

The judgment calls that cannot be taught. The instinct that something is off before you can articulate why. This only comes from years of focused immersion. It cannot be accelerated.

Context

Knowing how the industry actually works — who the key players are, what regulators actually care about, where the industry is heading, and what the unwritten rules are that never appear in any documentation.

Career arc

From the phones to EXCO to AI engineering.

1996–2005

Customer Services

Call centre → client liaison

Foundation in customer psychology

2005–2007

Momentum

Call Centre — Investments

92% client satisfaction

2007–2010

Momentum Wealth

Service Account Manager — LISP

95% satisfaction · 40% conversion

2010–2012

Sanlam SFA

Development & Sales Manager

16/18 vested · 23% revenue excess

2012–2014

Metropolitan Life

Branch Manager

94% budget adherence

2014–2015

Ithala Bank

LT Insurance Manager · Acting Div Head

R7.5M growth · FAIS 61%→86%

2016–2022

SOW Logistics

Operations & Business Dev Manager

20% on-time delivery improvement

2024–Now

CenturionAI

Founder & AI Engineer

Domain expertise → AI products

1996–2005

Where it started — customer services

It started at a call centre. MultiChoice, 1996. Not a managerial role — the phones. Customer psychology learned in real time, under pressure, with no margin for ambiguity. What people actually mean when they are frustrated. What makes a client stay. What makes them leave.

That was followed by customer service roles at Daimler Chrysler and others — each one adding commercial context, client relationship experience, and operational understanding. By the time financial services started in earnest in 2005, the foundation was already there. It does not show on the CV as a significant chapter. It shows in every client-facing system built today.

2005–2015

A decade inside South Africa's top financial services institutions

Momentum. Sanlam. Metropolitan. Ithala Bank.

That career did not begin at the top of financial services. It began answering investment queries in a Momentum call centre in 2005 — client and broker liaison for investment products. Technical understanding of investment instruments, asset classes, and transaction characteristics built from the ground up. 92% client satisfaction in an environment where a single processing error has direct financial consequences for someone else.

At Momentum Wealth Specialised Investments: managing large broker accounts — pension funds, provident funds, preservation funds, ETFs, retirement annuities, endowments, unit trusts, share trading. 95% client satisfaction. 40% proposal conversion rate. The kind of technical and relational depth that only comes from managing complex financial relationships over time.

At Sanlam SFA as Development and Sales Manager across four branches: not managing a team — building one. Recruited, trained, and vested 16 of 18 financial advisors. The team exceeded revenue targets by 23% in year one. That is a people development result. It is also a systems result — understanding which parts of advisor performance are trainable and which are structural.

At Metropolitan Life as Branch Manager: 94% budget adherence across full branch P&L — sales, operations, compliance, HR, client service. Running a branch means managing competing priorities simultaneously. The 94% number is not accidental — it reflects operational control over every lever in the building.

Then Ithala Bank. Long-Term Insurance Manager and Acting Divisional Head. Team of 21. Distribution across 49 branches. Development and management of a Long-Term Cell Captive — product design in collaboration with legal, reinsurance, actuarial, and finance teams. EXCO reporting. Prepared and delivered to Executive Committee.

And the FAIS Key Individual designation — RE1 and RE5, Categories I, II, IIA, III, IV — for the entire insurance division.

The Key Individual is the person the FSCA holds personally accountable for compliance outcomes. Not the company. The named individual. FAIS compliance was inherited at 61% and taken to 86% in twelve months. R7.5M revenue growth. 39% increase. 28% new business growth through targeted campaigns.

That is not a credential. It is accountability that was carried for real. It is why every AI system built for regulated industries here is designed from inside the compliance environment — not built to regulatory summaries, but built by someone who sat across from an FSCA auditor and owned the answers.

10 years

Inside SA financial services

RE1 + RE5

FAIS Key Individual

61%→86%

FAIS compliance in 12 months

R7.5M

Revenue growth, Ithala Bank

2016–2022

A different kind of domain expertise — operations and entrepreneurship

In 2016, a deliberate move into the private sector. SOW Logistics — Operations and Business Development Manager for nearly seven years. Full P&L. Cold outreach. Fleet management. Supply chain. Government contracts won through direct proposal development and executive-level relationships built from scratch in an industry that was entirely new.

On-time delivery improved by 20% in 12 months through route planning and real-time tracking. That is an operational result achieved without a budget increase — by someone who had to understand the logistics operation well enough to redesign it.

Seven years running a logistics SME teaches things that no corporate role teaches. What a payroll crisis actually feels like. What winning a government contract on cold outreach actually requires. What full P&L accountability means when it comes out of your own business — not a budget allocated by someone above you. Entrepreneurship as domain expertise: understanding what it means to run something, not just manage within it.

This chapter is often treated as a detour by people who do not understand it. It is not. Starting from zero in an unfamiliar environment, winning contracts, and running a full operation — that is the recurring theme in this career. AI was not the first time.

2024–Present

AI engineering applied to domain expertise

The decision to move into AI was not made because it was trending. It was made because the problems known best from thirty years of operational experience — manual compliance checks, slow document workflows, lead follow-up that nobody has time to do properly, financial services processes that depend on human memory — are exactly the problems agentic AI can solve. When built properly. By someone who understands the domain from the inside.

The no-code tools were tested first. Lovable. Make. n8n. Under real conditions, with real requirements, they showed their ceiling. The decision was to go deeper: Python, FastAPI, LangGraph, CrewAI, VAPI. Systems built to run when a client's business depends on them.

Four production systems live. A registered AI engineering studio. A full-time Product Manager — Sanele — who manages the client and product layer so the building does not stop.

The domain expertise is not a differentiator that gets mentioned in a pitch. It is the reason the products work the way they do.

The other half of the operation

Sanele — Product Manager

As CenturionAI grew, one gap became obvious. A serious operation cannot run on one person managing client relationships, product decisions, and technical builds at the same time. Something gives.

Sanele is CenturionAI's full-time Product Manager. She owns the client relationship layer — communication, requirement gathering, feedback loops, scope management, and making sure the product decisions being made are the right ones. She works directly with clients from discovery through delivery. She is not an assistant. She is the reason the operation runs the way it does.

The builder builds. The product manager manages. Both roles are full-time.

Education & credentials

BCom Marketing and Management

University of Pretoria

2007

FAIS Key Individual — Categories I, II, IIA, III, IV

FSCA

2014

RE1 Key Individual · RE5 Representative

FSCA

2012 / 2014

AWS Certified Cloud Practitioner

Amazon Web Services

2023

AI Engineer — Agentic Track (MCP & Agents)

Anthropic / Udemy

2026

AWS Solutions Architect – Associate

Amazon Web Services

In progress

Power BI Fundamentals

Exodus Experts

2024

Google Digital Marketing

Google

2022

The stack

PythonFastAPITypeScriptLangChainLangGraphCrewAIVAPISupabasePostgreSQLChromaDBPineconeAWSNext.jsPrismaLangSmithMCP

"Those that know me know I ain't afraid to get down and dirty. The risk is not building something new. The risk is staying with something broken because you've already invested in it."