Are We Racing to the Middle? A Developer's Take on AI

I started Vector back in 2018 with the goal of having a bit of side money to get me through my meager IT job. Since then, I’ve grown as a person and honed my skills. Midway through my journey, AI became a thing, and my company was not built around this new technology.

As a seasoned technologist (you were supposed to laugh at that), I wasn’t scared of the learning curve; I was, however, interested in the implications it would have on the market and public opinion. We now see this technology deployed at scale, becoming part of our daily conversations and a cornerstone for many companies.

The more I see agencies using this technology, the more I ask myself, “Are we just racing to the middle here?”

What is “The Middle?”

I don’t make that statement with any ego. If anything, the years I’ve invested in this career have shown me that I’m still barely scraping the surface of the development world. It humbles me.

What I mean by “the middle” is that point where you have just enough momentum to see a transformative work in progress, but not enough to get that surge of emotion that comes with fully understanding what is going on. It’s the zone of “good enough,” where speed can sometimes overshadow depth.

I am surprised by the breadth of knowledge that LLMs possess and how they are constantly evolving. But evolution isn’t always a linear path; while some things get better, other things get worse. For me, treading lightly is the best course of action.

A Framework for Sanity and Standards

This post isn’t meant to have everyone put down their coding agents. It’s an invitation to take a step back and ask if the real business value is more than just the end product. This is a reflection of my neuroticism about everything I do, which leads me to ask:

• “What did I learn using this tool?”
• “How can I implement this better in the future?”
• “Am I establishing a standard moving forward?”
• “How well do I understand what actually took place?”
• “Can I repeat these results?”
• “Can I troubleshoot this?”

To manage this, I’ve defined strict acceptance criteria for using LLMs in my workflow. Every project includes a structure to keep the AI accountable to my standards:

• 00-LLM.MD: A core set of information that establishes the ground rules. This includes the framework, libraries, naming conventions, etc.—the soul of the project.
• 01-designSpec.MD: A living document that tracks design decisions as they’re generated: fonts, border styles, backgrounds, and so on. It’s versioned, so I can see the history.
• 02-copy.MD: A detailed brief on everything related to the brand: copy, messaging, tone, target markets, SEO standards, and other marketing data.

This is how I’ve found a way to extract real business value from LLMs: by making them slaves to my standards and practices. It’s how I gain peace of mind.

The Journey is the Destination

Adopting this method allows us to invest time in more important things like client relations, infrastructure, marketing, blogging (he he), and other areas that require a less rigid approach.

Sure, I could just have an LLM write this for me, and to be honest, it would be less hassle. But I would feel like I was robbed of an opportunity to grow and get better at this. That thought permeates everything I do and how I approach life at large. It’s not just about the results, but more so about the journey, because ultimately, I need to feel like I’m moving at the same speed as my tools, not the other way around.

Beyond the Hype

I have seen companies use AI to take over entire business segments with a laser focus on results, without fully understanding the impact on their sales process, client relations, and quality. This creates endless wrappers for chatbots or aggressive API integrations that introduce more friction than they remove.

This all leads to the final part: the workforce. I tend to be an optimist, but I have seen the impact these tools can have. I know it can feel demoralizing for some people and has had a very real impact on how we think about work. I am hopeful that the bubble bursts and we swing the pendulum the other way—back towards stronger value propositions that take into account the human interaction in the work we do. That’s an aspect of business that is slipping away the more we see these tools rolled out.

To be clear, I’m not writing this to challenge anyone’s stance on LLMs. This is me sharing my thoughts on how these tools impact me and my brand.

Thanks for reading.