Software Development is Getting Weird
Software development is getting weird.
After more than 20 years in software development, Nate says his job has changed enough that even explaining what he does has become strange.
As he puts it:
“I mostly just talk to a computer now in English and make sure it doesn't go do anything crazy.”
At Scrum Day Madison on October 15, Nate will present “Software Development Is Getting Weird: How Do I Keep Up When AI Joins the Team?”
And he isn't talking about AI hypothetically. Nate and his colleagues at Lee Enterprises are already experimenting with AI throughout software development—and discovering that when AI removes one bottleneck, another one quickly appears.
When Writing Code Isn't the Bottleneck Anymore
For years, organizations have worked to improve the flow of software development. Now AI can generate code at a speed that completely changes that system.
Nate's team experienced it firsthand.
As more developers began using AI, they suddenly found their peer-review queue filling up. Nate recalls days when he spent virtually the entire day reviewing code. An AI agent might be capable of opening a dozen pull requests before lunch—but humans still have to determine whether that code is correct and whether it should have been created in the first place.
That raises a much bigger question: If writing code is no longer the constraint, what is?
Nate compares the shift to factories moving from steam power to electricity. The biggest gains didn't come from simply replacing steam with electricity. They came when manufacturers redesigned factories around what electricity made possible.
Nate believes software development may be facing a similar moment.
Instead of asking how we fit AI into our existing development process, he asks:
What would our process look like if writing code had always been easy?
AI Amplifies the Good—and the Bad
There's another catch: AI doesn't just make good teams faster.
It amplifies whatever is already there.
Strong engineering practices, clear requirements, cross-functional teams and fast feedback can become even more powerful. But weak refinement, missing scenarios, excessive handoffs and other organizational dysfunctions can accelerate too.
Nate's team has already seen examples: pull requests piling up, stories that aren't sufficiently refined for AI to work from, and missing scenarios and edge cases producing incorrect output at greater speed.
And AI has one significant advantage over the rest of us when something goes wrong: It doesn't have to live with the consequences.
As Nate points out, humans still do.
So, What Do We Do Differently?
That's what Nate will explore at Scrum Day Madison.
Attendees will look at how AI is changing refinement, development, testing and peer review; what belongs in the Definition of Done when agents perform some of the work; how teams can establish appropriate guardrails; and how Scrum fundamentals such as small increments, meaningful inspection and fast feedback become even more important as AI accelerates delivery.
And despite the seriousness of the subject, don't expect Nate to deliver a dry technology lecture.
He's been speaking publicly about software, Agile and decision-making since 2012, and his previous conference talks include “Your Brain is Broken and You Suck at Making Decisions” and “Teaching Agile to Your Grandmother, Your Five Year Old, and Your Manager.”
His philosophy is pretty simple: take the problem seriously, but have some fun while figuring it out.
Join Nate Adams at Scrum Day Madison on October 15 and discover what happens when AI joins the software development team.
