AI is changing how quickly software gets built. Engineers can generate code, troubleshoot problems and move from an idea to a working solution in a fraction of the time it once took. But as producing code gets easier, it raises another question for engineering teams: What separates someone who can make the code work from someone who understands what should be built, why it matters and what happens after it ships?
AI-generated code “clicks,” Argel Bejarano said. A problem that once took four hours to solve might now take 15 minutes.
The trouble is what that speed can hide. Engineers can plug in code because it functions, even when it doesn’t move the business forward and even when they don’t know the context behind the request.
“The code was never the problem,” Bejarano said.
Bejarano, Senior Staff Software Engineer, spoke with Developers News at droidcon 2026 in Orlando.
What separates code that works from code that adds value, he said, is the knowledge, judgment and communication that come with real seniority.
When the Product Doesn’t Click
Bejarano said many engineers don’t fully understand the product they’re building, which he called “a huge hiccup.”
The code may work, he said, “but then the product is not clicking in their mind.”
When that happens, a team can ship quickly and still miss the point. The feature exists, but nobody can say how it serves the customer or the company.
Knowing the Rules of Engineering
To show the difference between code that works and code that holds up, Bejarano described a colleague outside engineering who used AI to fix a bug. The fix worked, and Bejarano called the result impressive.
The colleague’s approach, as Bejarano summed it up: “It’s working. That’s it. Move on.”
What the colleague didn’t carry, Bejarano said, was the “burden of knowledge” that comes with engineering experience. An experienced engineer asks whether a change will be maintainable, whether it can be extended later and whether it follows established design principles.
“Is my code following the SOLID principles? Even that concern, he doesn’t have it,” Bejarano said, referring to a widely used set of object-oriented design guidelines.
Those questions are why the same fix would take Bejarano longer. They are also what keeps a quick fix from turning into a long-term cost for the business.
“Does that make him a senior? No,” he said.
What Seniority Really Means
For Bejarano, seniority isn’t measured by how much code someone can produce. He said engineers need to grow across several dimensions: technical skills, getting things done, product impact, communication and leadership. Increasingly, that also includes knowing how to use AI properly.
Communication may be the hardest to develop. Bejarano’s test is whether an engineer can explain the product to a family member from a different generation.
“If I’m thinking like in my mom or my dad, they are not as proficient as me,” he said.
If “the developer or the engineer cannot do that, then we have a communication issue that we need to solve,” Bejarano said.
It’s a standard Developers.Net co-founders Emilio Baez and Douglas Loyo, both engineers, also apply. If you can’t explain an idea to your grandmother, you have a communication problem.
The same thinking applies to hiring. Before asking for a title, he said, companies should ask what the work actually requires.
“Do I need a senior or do I need someone that is really good understanding the product?” Bejarano said.
AI can make code click faster than ever. Making sure that code is maintainable, understood and valuable to the business still depends on the engineers behind it.
As Bejarano put it, “the code has never been the problem.”
Developers.Net attends technology conferences across the country every month, connecting with the people scaling their businesses while building and implementing emerging technology. To connect with our team at an upcoming conference, email [email protected].
