We strapped a jet engine to a bicycle (SDLC and AI)

Just listened to the Syntax episode called WTF is a Software Factory? Man, did that ever hit at exactly the right time of things I'm dealing with as I'm realizing the entire way we write software is changing. Maybe that is obvious but if we were in hyperspeed before, now I feel like we hit ludicrous speed.

I don't want to rehash that episode but it got me to thinking about our SDLC (software development life cycle). Here is the analog I came up with (with Claude's help of course!)

We strapped a jet engine to a bicycle.

Our SDLC was built like a bike: human-powered, reliable, designed for human speed.

AI hasn’t given us a faster bike. It’s given us an engine. And bolting it onto the old frame doesn’t end well:

• Brakes fail. Testing, review, and security can’t stop what’s coming at them.
• Steering gets twitchy. Unclear requirements send you off the road faster than ever.
• The frame shakes apart. Architecture and integration buckle under the volume of change.
• The road can’t handle it. Pipelines and environments become the new bottleneck.

The destination hasn’t changed: deliver value to customers. But the vehicle has to.

And so does the driver’s role. Engineers move from pedaling to piloting: setting direction, monitoring, and making the judgment calls.

Stop asking how to make the bike go faster. Start asking what vehicle we need to build.

The next phase is up for grabs

It's one thing to evaluate how things aren't working but another to figure out the solution. I think the idea of software factories could be part of the answer but the future is still pretty fuzzy.

My primary goal is to give myself to this process. Digging into an old way of working won't last. And who knows, this next revolution of software development may actually be more fun, although the jury is still out on that one.