Five industries. Four countries. One pattern.

I started where most South African CAs start — in audit. At KPMG I spent three years pulling apart complex financial structures: 80-entity corporate groups, Investec and Nedbank lending books, cross-border restructurings. Led teams of 5–10 across Johannesburg and London. It taught me how businesses actually work — not the theory, but the plumbing.

A secondment at PwC in New York, then blockchain arrived. I saw what most accountants missed — not a speculative asset, but a new architecture for financial systems. I went all in. Built a community of investors and developers, and spent the 2017–2018 ICO boom advising clients on what to build and — more importantly — what not to. Most who ignored the advice regretted it within months.

At an energy technology company, I saw that countries with high smart meter penetration were sitting on untapped data value. I pivoted our product from energy trading to energy data, wrote the whitepaper, and got us into a competitive international accelerator in Australia. When my co-founder pulled out, I ran the entire program solo — architecture, product design, and daily investor pitches. COVID killed the traction, but the experience of building from zero changed how I work.

I joined an energy tech startup as the 4th employee and left as Financial Director of a 50-person company. In between, I designed a new finance product inside a major bank’s existing infrastructure — taking it from first conversation to first loan in six months and scaling the pipeline from 5 to 250+ clients per month. I built the financial model that became the engineering spec for the company’s automated platform. Automated operations so the business could scale without scaling headcount proportionally.

Today I build AI systems that do real work. At a $100M crypto fund, I automated my own role — replacing hours of daily manual research with AI pipelines I designed and built. Now, at Auto Alpha Advisory, I build autonomous AI agents that do the work, not just advise on it. The pattern across every chapter has been the same: find the bottleneck, prototype a fix, prove it works, automate it.