Building a Deterministic Environment… but Not for Computers
- 15 minsWritten by me, edited with robot assistance.
I recently gave the opening talk at Swift Bharat 2026 in Mumbai, India.
The title of my talk was:
?????????????????
Yes. That was the actual title.
I had spent MONTHS trying to figure out what I wanted to talk about.
I had ideas. Lots of ideas.
- Your App Needs a Harness
- My Brain Needs CI/CD
- Safe to Get It Wrong
- Just Enough Structure
- Protect the Fun
- Rediscovering Developer Tools
All reasonable topics. All things I genuinely care about.
I’ve been programming since 2004, building iOS apps since 2010, and giving technical talks for years. I’ve worked on fastlane, spent years helping developers ship software, organized Deep Dish Swift, and given more technical talks than I can probably remember.
This should have been easy.
But none of those ideas excited me. The more I tried to force myself into one of those topics, the more I realized that none of them felt particularly honest about the year I’d just had.
I wasn’t struggling to find a talk.
I was struggling.
And apparently it took me several months and a ridiculous number of slide revisions to figure that out. 😅
I’ve Never Been a Consistent Developer
For most of my career, I’ve had a pretty interesting relationship with productivity.
Most days I’m probably a 1x developer. Some days I feel like a 2x, 5x, or even 10x developer. And then there are the 0x days, where getting anything meaningful done feels almost impossible.
Historically, this worked out fine. I’d have a few bad days followed by some REALLY good days, and everything would balance out. I could brute force my way through difficult periods or make up for them later when I had the energy and motivation.
I built an entire career around this style of working. And I was pretty successful with it!
Until 2025.
Suddenly, the bad days weren’t something I could recover from with a few good ones. I couldn’t brute force my way through them anymore, and the working style I’d depended on for two decades was becoming unsustainable.
That was terrifying because I didn’t really know another way to operate.
I’ve written about parts of this before in Understanding My Newly Recognized Operating System, but here’s the abbreviated version.
In early 2025, I was becoming increasingly overstimulated. Sounds, lights, conversations, context switching, and all the little things I’d somehow managed for most of my life were becoming harder and harder to deal with. I was masking heavily, pushing myself to be the person I thought I needed to be, and spending an enormous amount of energy doing things that seemed effortless for other people.
In October 2025, I was diagnosed with anxiety. I questioned that diagnosis because it didn’t really match how I understood myself, but I was struggling and wanted help, so I accepted it and tried the treatment.
The medication made some of the overstimulation quieter. Unfortunately, it also made just about everything else quieter. Including the things I enjoyed. Including building software.
By the end of 2025 and into early 2026, I was experiencing burnout unlike anything I’d dealt with before. I spent two weeks on the couch and took an additional week off work. Not building things, not playing with side projects, and not doing the things that usually make me happy. Just existing.
Which is VERY unlike me.
Eventually, I went through a reevaluation, and in May 2026 I was diagnosed with autism and ADHD. AuDHD.
Suddenly, a lot of things started making sense. The routines, overstimulation, need for predictability, hyperfocus, and wildly inconsistent productivity weren’t random quirks. There were explanations for things I’d been experiencing my entire life.
It wasn’t that I’d suddenly become a different person. I’d spent decades operating under the wrong assumptions about how my brain worked.
I finally had some answers, and I was relieved.
But I wasn’t magically better.
Wait… I Don’t Know How to Do This Anymore?
One of the more surprising parts of this whole experience was skill regression.
Things I’d been doing for YEARS suddenly felt difficult. Tasks that used to happen almost automatically required conscious effort, and I found myself struggling with things I’d previously considered second nature.
I’d spent two decades becoming good at building software. My ability to create things was a huge part of how I understood myself.
And now I was questioning whether I could even do it anymore.
Or, more uncomfortably…
Should I keep doing software?
That’s a pretty wild question to find yourself asking after spending most of your life doing something, especially when that something has been such a big part of your identity.
But eventually I started realizing that software wasn’t necessarily the problem. The way I’d been operating was.
I’d built my career around being able to compensate for bad days with exceptional ones. I depended on those 2x, 5x, and 10x days.
But what happens when those days don’t come? What happens when you can’t brute force your way back?
Maybe the answer wasn’t to try harder.
Maybe I needed to change the environment I was operating in.
Understanding the System
One of the most interesting things about my autism and ADHD diagnosis is how seemingly contradictory the two can feel.
My autistic brain really likes predictability, fewer decisions, and low stimulation.
My ADHD brain really likes novelty, momentum, and feedback.
These things don’t always play nicely together.
And for most of my life, I didn’t even realize I was trying to balance them. I was just doing whatever seemed to work. Sometimes that meant building elaborate routines, and sometimes it meant jumping into a brand-new project at 11 PM because I had an idea and absolutely HAD to see it through.
Sometimes it meant both.
But after my diagnosis, I started looking at things differently.
Instead of trying to force myself into being more productive, I started asking what my brain actually needed to operate consistently.
And I started making changes.
Some of them were surprisingly boring. I wear a lot of the same clothes, eat a pretty simple and predictable diet, break exercise into smaller workouts, and use hats, earplugs, and headphones to control stimulation.
And perhaps most surprisingly…
I started doing fewer side projects.
I KNOW. That last one is a pretty big change for me. 😅
But these changes helped. A LOT.
I was spending less energy making decisions, dealing with overstimulation, and trying to force myself to operate in environments that weren’t working for me.
And slowly, over several months, I started feeling something I don’t think I’d ever really experienced before.
Consistency.
Not the occasional incredible day followed by several terrible ones. Just consistently okay days.
And honestly? That’s been pretty amazing.
I had no idea how much of my energy was being consumed by simply managing my environment. By making parts of my day more predictable, I was leaving more room for the things that actually mattered.
My day was becoming more scripted.
And somehow, that was giving me more freedom.
Dear Josh, This Is a Developer Conference
Okay. At this point in my talk, I had spent a significant amount of time discussing my brain, my routines, and my personal struggles.
Which is probably not what people expected from someone who has spent years talking about developer tooling.
I even had a slide reminding myself that I was, in fact, speaking at a developer conference.
Don’t worry. We’re getting there.
Because while I was trying to understand my brain, something else was happening.
Something that was fundamentally changing my relationship with building software.
AI.
I’ve Never Felt More Powerful
AI has made it easier than ever for me to build things.
I can have an idea, open a terminal, describe what I want, and start seeing results almost immediately. Things that used to take hours can happen in minutes, and things I might have abandoned because they required too much setup suddenly feel accessible.
And for someone who LOVES building things? That’s incredible.
I’ve never felt more powerful as a developer.
But I’ve also never felt more dangerous.
And I’ve never felt more out of control.
Because the barrier between having an idea and acting on it has almost completely disappeared.
I can just keep going. And going. And going.
Vibing. Token maxing. Building things I could build but probably shouldn’t.
Generating more code than I can reasonably review. Creating more things than I can realistically maintain. And sometimes doing all of that without really understanding what happened along the way.
That’s a weird feeling for someone who has spent most of their career caring deeply about understanding how things work.
I’ve always loved building things, but now I could build things faster than I could meaningfully process them.
And I started wondering whether that was actually helping me.
AI Was Feeding One Part of My Brain and Starving Another
Remember those two sets of needs?
Autism: Predictability. Fewer decisions. Low stimulation.
ADHD: Novelty. Momentum. Feedback.
Well…
AI is REALLY good at feeding the ADHD side.
There’s always something new. Always another prompt, another feature, another thing to try. The feedback is immediate, and the momentum can feel endless.
And that can be incredibly exciting.
But my autistic need for predictability? That wasn’t necessarily being met.
I was creating more uncertainty, more things to review, more things to understand, more decisions, and more unfinished work.
I was getting all the novelty and momentum I could possibly want while losing the stability I’d spent months trying to build.
And I think that combination contributed to my burnout and skill regression.
I don’t think AI was the only cause. There were obviously a lot of things happening in my life at the same time. But I do think it amplified some of the patterns that were already causing me problems.
The funny thing is that I’d spent years trying to make my life and work more predictable, and then I found myself enthusiastically using something that could make both feel less predictable than ever.
Oops.
Struggle Is Really Real
So what was this talk actually about?
Struggle.
Struggle is real. Struggle is REALLY real.
And despite spending months understanding myself, making changes, and finding more consistency, I still struggle.
In fact, I was struggling with the talk itself.
The night before Swift Bharat, I was having a terrible time preparing. I’ve given plenty of talks. I’ve spoken in front of large audiences. I organize a conference. This should be something I know how to do.
But that night, it felt like I’d never given a talk before.
Everything felt new and different. Things that used to come naturally weren’t coming naturally anymore, and I was incredibly frustrated.
On the drive to the conference, Charlie Chapman asked me how my preparation had gone the night before. Charlie and I are almost always joking around with each other, so you’d probably expect me to give him some ridiculous answer.
But this time I was completely serious.
Terrible.
That was my answer.
And I probably shouldn’t have been surprised.
I’d spent months talking about skill regression, and here I was experiencing it while preparing a talk about skill regression.
Pretty on-brand, honestly. 😅
But I was there. I was giving the talk. And I was doing something that had felt incredibly difficult just the night before.
I don’t think that means I’ve figured everything out. Some skills still don’t feel the way they used to, some days are still difficult, and I’m still learning how to work with a brain that I spent decades misunderstanding.
But maybe that’s okay.
Maybe I don’t need to have everything figured out to keep going.
The Ground Is Changing Under Us
Everything I’ve talked about so far is my experience.
My brain. My diagnosis. My burnout. My relationship with building software and AI.
I know not everyone will relate to those specific experiences, but I also know I’m not the only person struggling.
And I think more people are going to find themselves asking some of the same questions I’ve been asking.
Because the ground is changing underneath us.
The way we build software is changing. The skills we rely on are changing. And the things that made us successful before might not work the same way anymore.
AI is changing our industry incredibly quickly, and I don’t think we fully understand where all of this is going.
I certainly don’t.
For me, that uncertainty arrived at the same time I was questioning almost everything about how I work. I was trying to understand a brain I’d spent decades misunderstanding while the industry I’d spent my career in was changing around me.
But I think there’s something important in that experience that extends beyond autism, ADHD, or burnout.
I spent the last year learning about myself. I had to question assumptions I’d been operating under for two decades, figure out what was helping me, identify what was holding me back, and make changes to an environment that wasn’t working anymore.
And now I think we’re all going to have to ask some of those questions.
Not just about our tools.
About ourselves.
What helps you thrive? What’s holding you back? What needs to change?
Those aren’t necessarily questions with easy answers. They certainly weren’t for me. And I don’t think they’re questions we can answer once and never revisit.
The things that made us successful five years ago might not be the things that make us successful five years from now.
That doesn’t mean everything we’ve learned is suddenly worthless. It doesn’t mean we need to abandon the things we love, and it doesn’t mean we need to blindly embrace every new tool or workflow that comes along.
But maybe it does mean we need to pay closer attention to ourselves.
What are we doing because it genuinely works for us? What are we doing because it’s how we’ve always done things? And what happens when those things stop working?
I spent most of my career assuming I understood how I worked best.
Turns out I was missing some pretty important documentation.
And now that I have a better understanding of myself, I’m realizing how many things I can approach differently.
Not necessarily better. Not necessarily faster.
But more sustainably.
And hopefully with a lot more enjoyment along the way.
Building a Deterministic Environment
I spent months trying to figure out what my Swift Bharat talk was about.
I thought I had seven different ideas. I thought I needed to pick one. I thought I needed a technical topic with a clear lesson and a satisfying conclusion.
But none of those ideas felt right because they weren’t really addressing what I’d been experiencing.
I was struggling, and I needed to understand why.
I needed to stop assuming that the way I’d always worked was the way I had to keep working. I needed to understand what helped me thrive, what was holding me back, and what I could change to make things more sustainable.
And that’s when the original idea of building a deterministic environment started making sense to me in a completely different way.
I’d spent years thinking about development environments, tooling, and workflows. But the environment I really needed to understand was my own.
Not my Mac. Not my terminal. Not my CI/CD pipeline.
Me.
So it turns out I DID have a title for my talk.
Building a Deterministic Environment
Start with understanding yourself.
I don’t know exactly what the future of software development looks like. I don’t know what my own relationship with building software will look like in a few years, and I don’t know whether I’ll ever feel quite like the developer I used to be.
But I’m also starting to realize that maybe I don’t need to.
I’m not trying to rebuild the exact version of myself that worked before. I’m trying to build an environment that works for the person I actually am.
And with everything changing around us all the time, maybe the most important thing we can do is stay tuned to ourselves.
That’s where I’m at.
Still struggling. Still building. Still figuring it out.
And hopefully still having fun. 😅