The Return of the Craft

Welcome to Graphish!

Here’s where I share short observations, discoveries and learnings- they’re opinionated but authentic. My experience is best viewed through the lens of an engineer, product builder, and team champion in small to medium organisations, deploying reliable, secure, mission-critical applications into production in enterprises and sensitive large entities. This is important to note as the dual-edged sword of freedom and constraints in this environment shapes my thoughts.

So, what’s been occupying my mind?

Of late, I’ve been exploring whatever everyone is exploring - AI code assistants and agents- to get firsthand experience of whether the “Holy shit, I just built <insert groundbreaking idea> overnight” moments are really worth invoking the divine. Stay tuned to future posts to read about my feelings and observations on the topic. For now, my gut instinct, based on my own experiments as well those of people around me, is a happy and hopeful one. I’ll go out on a limb to say:

Engineering craft will return.

It feels like a distant memory, when writing code also meant solid engineering and thinking really hard about how to work with the resources you had- or rather, didn’t have: compute and memory. Then we went through a period of abundance, where, at least for general-purpose business applications, you could get away with some degree of sloppiness. Throwing tools/frameworks/money at the problem sort of worked. And now we’re at the stage where one can spend even less time building applications often without consciously or purposefully engineering anything.

So is this a bad thing? Are some of us just clinging to the glory days of crafting solutions? I think not.

Along with the seductive pull of coding agents comes the cost, whether explicit or not. Cost is the new precious resource that small to medium businesses and SaaS providers will need to optimise for. Not just cost incurred due to the very architecture and design of the application, but also from maintainability, security (or the lack of it), fault tolerance, infrastructure demands, and so on.

This is one of several reasons why I believe the craft must return. Not in the form of writing compact and clever code, but in designing systems that keep costs down, and in the experience and skill to recognise and improve inefficiencies. We’ll be back to assessing and managing risks and deciding on trade-offs- an agent or hand written code for a specific task? Quality versus cost while selecting a model? And figuring out creative ways to squeeze out as much as you can without burning the budget of a tiny island nation.

I’d expect to see cost planning become a more upfront consideration. Variable costs like the ones we’re faced with today will inevitably put pressure on the bottom line and not having a grip on these costs will also lead to poorly informed outcome based pricing models.

As an engineer, I welcome the return to solid fundamentals, to the deep satisfaction of solving a problem, and the joy of crafting products with care. Not in a snobbish way but in ways where we think through the consequences or deliberately choose the boring tech when it’s the right thing to do, where you have hammers of different shapes and sizes but it is clear that not everything is a nail.

Of course, the question still on everyone’s mind is: will we lose our jobs? I’d ask a different one- will software engineering as a profession continue to draw the droves of developers it’s managed to attract over the last decade or so? Or will the engineering community shrink back down to those who truly value and enjoy the craft, finding innovative and satisfying ways to work with AI. Only time will tell.

Recommended sound track: Back in Black (AC/DC)