The Software Paradox

There’s a tension that often gets overlooked in the hype around LLMs and agentic development: code is both an asset and a liability.

Software drives competitive advantage — proprietary code can be worth billions, powering products, platforms, and processes that differentiate a business. Yet every line of code also comes with a cost: it must be maintained, updated, and secured.

Over time, even the most valuable software can become a liability if neglected, accumulating technical debt and creating dependencies. The same code that generates value today can create risk tomorrow.

This paradox has always existed, but now, with AI generating thousands of lines of code from a single prompt, vigilance is more critical than ever.

The engineering practices born from past project failures matter even more now. The hope is that AI will eventually internalise these lessons some how and refactor the slop that earlier versions produced.

But if that hope fails to materialise, many businesses will be burdened with unmaintainable, risky, and expensive code - undermining the very advantages they hoped AI would deliver.