The Average Developer is a Multitasker: A Case for Agents
I’ve always been a proponent of coding with AI, but I used to be wary of pushing the “productivity” and “fully automated workflow” angles. For me, the real benefits were accessibility, psychological safety, and education. I didn’t want to encourage executives to demand unrealistic output from developers or, worse, push them out of their jobs. But as AI tools have evolved from tab autocompletion to agentic workflows, I've changed my mind. AI can make developers more productive—just not in the 10x/rockstart/ninja developer way we typically imagine. The average developer is a parent. I’m reminded of this daily as I’m on parental leave. Coding and blogging is relaxing for me. However, while my newborn is (slowly) developing a circadian rhythm, I’m finding it tough to write or code for fun. Case in point: this entire blog post is being composed on my phone while I nurse. Kids get sick, schedules shift, and parents are constantly pulled away from the keyboard. There’s a ton of context switching, and when you’re coding, each interruption can cause mistakes or lead to forgetting key details. Many AI agents have a memory feature that can keep track of and summarize the work completed. And maybe while a busy parent is tending to their sick child, an AI agent can kick off and manage long-running tasks like a database migration. This enables developers to show up as valuable teammates and present parents. The average developer is not a developer. Many people who write code hold product designer, engineering manager, or product manager titles. Because they wear multiple hats, they need to rapidly validate ideas through proof-of-concepts. Instead of spending hours creating mockups, they can use AI agents to build full stack prototypes. This gives them more room to focus on high-level strategy and problem solving. Check out this video where Addy talks about using agents to build a full stack app with authentication. The average developer is a teacher. Some developers are bootcamp instructors or professors who grade assignments and answer student emails late into the night. An AI agent can help by generating visual demos (like a quick bubble sort animation) for the next day’s lesson, leaving instructors more time to connect directly with students or refine course materials. The average developer is an open source maintainer. They might be a full-time software engineer by day and unpaid (or lightly compensated) open source maintainer by night. Users depend on frequent updates, but sometimes the maintainer just needs an evening off. An AI agent can triage issues and initiate automated deploys—helping maintainers sustain momentum in the community without sacrificing all their personal time. Agents aren’t perfect, but they’re helpful. AI tools—even the ones marketed as having “PhD-level intelligence”—still get things wrong, so I don’t believe agents should or even could replace developers. However, I believe that today’s developers are multihyphenates often juggling additional roles like founder, freelancer, traveling developer advocate, and meetup organizer. And we need a sidekick to help us parallelize tasks, keep tabs on project details, maintain context across interruptions, and support the multithreaded nature of our day. My interpretation of AI enabling developer productivity is not focused on cranking out more code. Instead, I mean that it can help developers reclaim their time—whether that’s caring for family, exploring side projects, or just taking a breather. The average developer is a multitasker and our tools should reflect that.
I’ve always been a proponent of coding with AI, but I used to be wary of pushing the “productivity” and “fully automated workflow” angles. For me, the real benefits were accessibility, psychological safety, and education. I didn’t want to encourage executives to demand unrealistic output from developers or, worse, push them out of their jobs.
But as AI tools have evolved from tab autocompletion to agentic workflows, I've changed my mind. AI can make developers more productive—just not in the 10x/rockstart/ninja developer way we typically imagine.
The average developer is a parent.
I’m reminded of this daily as I’m on parental leave. Coding and blogging is relaxing for me. However, while my newborn is (slowly) developing a circadian rhythm, I’m finding it tough to write or code for fun. Case in point: this entire blog post is being composed on my phone while I nurse.
Kids get sick, schedules shift, and parents are constantly pulled away from the keyboard. There’s a ton of context switching, and when you’re coding, each interruption can cause mistakes or lead to forgetting key details.
Many AI agents have a memory feature that can keep track of and summarize the work completed. And maybe while a busy parent is tending to their sick child, an AI agent can kick off and manage long-running tasks like a database migration. This enables developers to show up as valuable teammates and present parents.
The average developer is not a developer.
Many people who write code hold product designer, engineering manager, or product manager titles. Because they wear multiple hats, they need to rapidly validate ideas through proof-of-concepts. Instead of spending hours creating mockups, they can use AI agents to build full stack prototypes. This gives them more room to focus on high-level strategy and problem solving.
Check out this video where Addy talks about using agents to build a full stack app with authentication.
The average developer is a teacher.
Some developers are bootcamp instructors or professors who grade assignments and answer student emails late into the night. An AI agent can help by generating visual demos (like a quick bubble sort animation) for the next day’s lesson, leaving instructors more time to connect directly with students or refine course materials.
The average developer is an open source maintainer.
They might be a full-time software engineer by day and unpaid (or lightly compensated) open source maintainer by night. Users depend on frequent updates, but sometimes the maintainer just needs an evening off. An AI agent can triage issues and initiate automated deploys—helping maintainers sustain momentum in the community without sacrificing all their personal time.
Agents aren’t perfect, but they’re helpful.
AI tools—even the ones marketed as having “PhD-level intelligence”—still get things wrong, so I don’t believe agents should or even could replace developers.
However, I believe that today’s developers are multihyphenates often juggling additional roles like founder, freelancer, traveling developer advocate, and meetup organizer. And we need a sidekick to help us parallelize tasks, keep tabs on project details, maintain context across interruptions, and support the multithreaded nature of our day.
My interpretation of AI enabling developer productivity is not focused on cranking out more code. Instead, I mean that it can help developers reclaim their time—whether that’s caring for family, exploring side projects, or just taking a breather.
The average developer is a multitasker and our tools should reflect that.
What's Your Reaction?
