The Pomodoro timer was still running when I opened the fourth browser tab about something completely unrelated. Twenty-three minutes left on the clock. Search architecture document still open. We were adding personalized recommendation search through AI ... the kind of deep work that requires holding complex state in your head. The productivity system hadn't failed. I had outgrown the container it assumed I could fit into.
The productivity industrial complex has a formula. Block your calendar. Close your notifications. Set a timer. Focus.
I tried it all. The 25-minute intervals. The deep work blocks. The morning routine optimization. Each system promised the same thing: if I just arranged my environment correctly, my brain would cooperate.
The problem was never the environment. The problem was that these systems were designed for brains that naturally return to tasks.
ADHD isn't a focus problem. It's a neurological difference in executive function ... planning, prioritization, motivation, impulse control. Research on sustained attention in adults with ADHD shows that performance deteriorates significantly over time (what researchers call "time-on-task effects"). When tested on 20-minute sustained attention tasks, adults with ADHD showed moderate deficits in alertness, selective attention, and divided attention compared to neurotypical controls. The structured work periods that productivity systems promise don't reset attention for neurodivergent brains the same way they do for neurotypical ones. (Source)
We can hyper-focus on things that interest us, but we struggle to sustain focus on demand. We can start strong, pull focus for maybe a day, but our brains don't naturally come back to the thing we were doing. Not because we lack discipline. Because our wiring doesn't loop back the way neurotypical brains do.
Not because we lack discipline. Because our wiring doesn't loop back.
That search feature was complex. Personalized recommendations through AI required holding vector embeddings, ranking logic, and platform integration patterns in working memory all at once. The kind of work where you need to hold architecture in your head while you build. I'd get started, make progress, then get pulled sideways. Not by distraction ... by the natural pattern of a brain that moves faster than its container allows.
The productivity advice assumed the bottleneck was external. If I could just eliminate interruptions, I'd focus. But my interruptions were internal. Five solutions generated before my hands could type one. Architecture reconstructed three times while trying to write the first function. The gap between ideation speed and execution speed created a traffic jam of unexpressed thinking.
The gap between ideation speed and execution speed created a traffic jam.
The Realization
The breakthrough wasn't another productivity system. I just stopped believing I needed to focus harder.
I could sustain attention. The workflow just assumed attention should be sustained through mechanical execution. Typing one file, one function, one line at a time, while my brain had already moved to the fifth iteration.
What I needed wasn't better discipline. It was different mechanics.
I found them building a CLI tool for DevOps automation.
The Redesign
I was building a CLI tool for DevOps automation. Started with a base command. Next thought: write the tests. Then: concurrency issues might slow this down, make it async. Then: API calls need rate limiting. Then: circuit breakers, retries.
With the old workflow, I'd have typed one file, gotten pulled into something else, lost the thread, and three hours later had half a command with four better approaches forgotten.
With AI, I described the base command. Claude wrote the first pass. It used any for everything. I pushed back. We're using TypeScript, not JavaScript with extra steps. It tightened the types. I described the test pattern. It wrote the tests. I described the async pattern, the rate limiting, the circuit breakers. Each iteration took minutes, not hours.
The five solutions in my head didn't decay in the queue waiting for my fingers. They became the next five prompts.
The five solutions don't decay waiting for my fingers. They become the next five prompts.
This is amplification, not delegation. I'm still making every architectural decision. Still reviewing every abstraction. The difference is I'm working at the speed I think instead of the speed I type.
Which is when you see the criticism for what it is. Go to r/ExperiencedDevs. Scroll any AI post. You'll find the mantra: "AI is making engineers lazy."
The angle they miss is neurodivergence. This isn't about AI making a certain type of person lazy. That person was already lazy.
Engineers are inherently lazy. That's the job. The best engineers automate. They find cheat codes. They optimize. Now we're classifying which lazy we approve of and which we don't, which is its own irony.
I'm lazy. When I'm engineering, I take the easiest approach because I have twenty things to do and each one needs my brain. If I can make them faster, I'm not thinking less. I'm thinking more deeply.
That's something.
The grind was never proof of skill.
I spent fifteen years trying to fix my brain to match the tools. The tools should fit the brain. That's not limitation. That's design.
What This Means for Leaders
Look at your team. Someone is holding five solutions in their head and losing four of them to the gap between thinking and typing. They're not distracted. They're not undisciplined. They're waiting for a workflow that matches their speed.
The engineers who think fastest often ship slowest. Not because they lack skill. Because the system is built for hands that move at average speed.
AI doesn't just change what your team builds. It changes who can build at their natural pace. That's worth noticing.
Now everyone gets to find out.
One email a week from The Builder's Leader. The frameworks, the blind spots, and the conversations most leaders avoid. Subscribe for free.
Top comments (0)