By André van der Walt, Solutions & Software Engineer at Insight Consulting. Blog post 2 of 4 in the Prompted series.
So often, my team has sat in conversations with clients where they went looking for the “proper” system they needed for a given process (HR, Finance, Project Management etc.) only to find that the pricing is way out of budget. Many of the “industry standard” solutions to common problems are built, priced, and marketed for the US, UK, or EU markets. At an exchange rate so often on the wrong side of R16 to the Dollar, those sums don’t make sense for a cost-sensitive South African market. So, we “stick with a spreadsheet for now” and sacrifice many hours to running processes by hand, while the tool sits so visibly out of reach.
The build-versus-buy maths has always been interesting in the South African market. The benefits of established solutions are blatantly obvious and so easy to justify before the quote lands. Then we so often pivot to cheaper options, start accepting more compromises and bend our requirements to what the tool can do. The final option is to build it (or rather, have it built). With the right partner we end up with a bespoke solution that fits our business, but the costs are also the most variable of the three options. AI has changed the order in which these options are considered.
The acceleration of output from developers due to AI has translated directly into bespoke solutions deployed faster. They ship with the right features catered to the specific business for which they were built. And fewer hours are wasted in requirement gathering, testing or refinement. The result is that an off-the-shelf solution that was built to service tens of thousands of companies loses its lustre. Why would a business want to adopt new nomenclature and processes with the need for extensive training on a mammoth piece of software? Especially when the payoff only materialises sometime in the future. All the while, deeply entrenched employees are asking: “What was wrong with how we have been doing that? I understood that process”.
The catch, death, and taxes. Of course there is a catch, or else this would all be too good to be true and everyone would just employ a vibe coder and bake all their software in-house. Unfortunately, the truth is that, while it did reduce, the cost migrated elsewhere. AI-enabled devs had to learn a new skill: Token Management.
Devs have more options than are sensible for tooling and AI subscriptions. The subscriptions all start free, then $20 per month, then $40, $60, $100, $200, extra usage ad infinitum. The first bite of productivity and deferred work leads quickly to running multiple coding agents in parallel: researching, coding, documenting. This cost balloons quickly, and without the discipline and skill to rein it in and recognise errant paths, you end up wasting more money than the value you create.
Token Management comes with more nuance than just saying “use a cheaper model” because the fact of the matter is that there are times where that will actively harm the objective. It comprises hard skills like context engineering, prompt engineering and use of tools and harnesses but it also needs soft skills like good judgement, diligence, and knowledge through experience. These soft skills are hard-won through exposure, experience, and time – things AI does well, but not quite well enough. When you exercise all these things together, you see the benefits of accelerated development, reduced wastage, and improved solutions – exactly what makes a bespoke build worth it in a cost-sensitive market.
It’s a shift I think we’ll see more of in the coming years. More spreadsheets will get replaced with custom solutions instead of waiting to hit the point at which the off-the-shelf solution makes sense. Of course, the big “industry standard” solutions will continue to exist; they are too broadly used to disappear. But fit for purpose systems are no longer the preserve of those with bottomless wallets. For now, it’s worth asking which spreadsheets in your own business you could finally replace with the proper tool.
This article first appeared in Insight Edge, our newsletter on LinkedIn.