The Pace of Possibility: AI Development is Amazing, and Here's What Keeping Up Actually Feels Like
Let me be clear: AI-assisted development is extraordinary. Not hype extraordinary. Not “this will change everything someday” extraordinary. Actually, tangibly, right-now extraordinary.
I built five production applications in three months. A train tracking app during a two-hour delayed journey. A full venue management system that now runs payroll for a local business. A horse gait analysis app that processes sensor data in real-time. A collaborative meal planning app that replaced two paid subscriptions. A UK fuel price finder with CarPlay integration. All while working full-time. All shipped to real users.
This isn’t theory. This is happening. The possibilities are genuinely enormous.
And the practical reality of trying to keep up with those possibilities? That’s what I want to talk about.
The Possibility Explosion
Here’s what’s actually possible now: you can have an idea at breakfast and a working prototype by lunch. Not a sketch. Not a mockup. A working prototype.
Before Claude Code, I had a mental backlog of projects I’d “build someday”:
- Train tracking with offline caching
- Venue management with arena bookings
- Gait detection using phone sensors
- Probably a dozen others
These weren’t vague dreams. I had architecture sketches. I understood the APIs. I knew what the MVP would look like. What I didn’t have was time. Building each would take months. And months turn into years. And years turn into “that thing I thought about once.”
Now? I open Claude Code, describe what I’m building, and it builds it. Not perfectly. With iteration. With refinement. But it builds it. The gap between “I want this to exist” and “this exists” has collapsed from months to days.
This is the enormous possibility everyone talks about. And it’s real. It’s not coming. It’s here.
What’s Actually Possible: Real Examples
LiveRail: Two hours on a delayed train. From “I’m annoyed at juggling five apps” to “working train tracking app with offline caching.” By the weekend, it was on the App Store. With Live Activities. CarPlay support. Multi-language translations. This would have been a 6-month project before. It took a weekend.
Equestrian Venue Manager: Started as a simple booking system to replace an expired WordPress site. Within weeks: arena bookings, horse management, health tracking, turnout planning. By Christmas: full HR system with timesheets, leave management, payroll integration. Last month, real people got real paychecks calculated by this system. People depend on it.
TrackRide: Horse gait detection using phone sensors. FFT analysis. Hidden Markov Models for state transitions. Biomechanical calculations. Live Activities showing gait transitions. This is hard stuff. It exists. It works. It processes real sensor data from real rides.
TableTogether: Frustrated with juggling Paprika 3 and MyFitnessPal, I built a unified meal planning app in a month. Recipe catalogue, collaborative household planning, automated grocery lists, personal macro tracking with natural language calorie estimation. SwiftUI, SwiftData, iCloud sync. Our household uses it daily. Two paid apps replaced with one custom solution.
FuelFinder: UK fuel price finder for iPhone and CarPlay. Plan a route, discover the cheapest fuel stations along your journey. Corridor search algorithm, OAuth2 integration with UK Government Fuel Finder API, Core Data with iCloud sync for favourites. Built in days, not months. Real-world route planning with price optimization.
These aren’t demos. These aren’t prototypes gathering dust. These are production applications solving real problems for real users. And I built them while working full-time, having a family, maintaining other commitments.
The productivity gain isn’t theoretical. It’s measurable. It’s real. This is why everyone is excited about AI-assisted development.
But here’s what nobody talks about enough: what it actually takes to maintain this pace.
The Energy Vampire
Steve Yegge wrote about The AI Vampire in February. His argument is straightforward: AI makes you 10x more productive, but it drains your energy like the vampire character from “What We Do In The Shadows.” You produce more. You accomplish more. And you’re exhausted.
He describes “nap attacks” - falling asleep suddenly after long AI coding sessions. His company seriously considered installing nap pods. The fatigue is real.
I get it now.
It’s not physical tiredness. I’m not lifting heavy things or running marathons. It’s the mental exhaustion of thinking at full capacity for hours without break.
With traditional development, there are natural pauses. You hit a problem. You search Stack Overflow. You read documentation. You try something, it doesn’t work, you walk away for a bit. These pauses are frustrating, but they’re also rest.
With Claude Code, the pauses disappear. You describe a problem. It suggests a solution. You review it. You suggest modifications. It implements them. You test. You find an edge case. You describe it. It fixes it. The loop is continuous.
And continuous high-level thinking is exhausting in ways that writing code never was.
The Relentlessness of Always-On Capability
The hardest part isn’t the work. It’s that the work never has to stop.
Before, I’d work on a project until I hit a blocker. Missing library. Unclear API documentation. Architecture decision I didn’t know how to make. These blockers forced breaks. You’d step away, think about it, come back later.
Now, Claude Code either solves the blocker or helps you think through it. There’s always a next step. Always progress available. The only reason to stop is choosing to stop.
And choosing to stop feels wrong when you’re in flow. When the agent is right there. When one more feature is “just 30 minutes away” (it’s never just 30 minutes, but it feels that way).
I find myself thinking about projects constantly. Not because I have to. Because I can. Because the gap between “I wonder if…” and “let me try…” has shrunk to seconds.
This is the vampire effect. Not that AI drains you while you use it. That it makes creation so accessible, so effortless-feeling, that you never want to stop. And not stopping is exhausting.
The Math Doesn’t Add Up
Yegge’s other point hits hard: if AI makes you 10x more productive, you’re producing nine additional engineers’ worth of value. Your employer captures 100% of that value. You get… nothing. Or at least, not 9x your salary.
I’ve built five production applications in three months in my spare time. These aren’t weekend hacks. LiveRail is on the App Store. EVM processes real payroll. TrackRide analyzes actual horse gaits. TableTogether replaced two paid subscriptions. FuelFinder optimizes fuel costs on road trips.
What’s the economic value of this productivity? I genuinely don’t know. The apps exist because I wanted them to exist, not because I’m trying to monetize them. But the question lingers: if I can build five production apps in three months part-time, what does that mean for my day job? For employment? For the industry?
I don’t have answers. Just questions I think about at 11pm when I should be asleep.
The Happy Exhaustion Paradox
Here’s the strange part: I’m happier than I’ve been in years.
I’m building things I’ve wanted to build for a decade. Problems I’ve mentally sketched solutions for finally have implementations. Ideas don’t languish in the “someday” folder anymore.
The creative satisfaction is real. Seeing LiveRail work offline in a tunnel. Watching EVM calculate payroll correctly. Getting the gait detection algorithm to distinguish walk from trot. These are wins. Tangible accomplishments. Things that work and solve problems.
But I’m also perpetually tired. Not burned out tired. Not “I hate this” tired. Just… depleted. Like I’ve been thinking hard for months without pause. Which I have.
It’s a strange combination. Happy exhaustion. Fulfilled fatigue. The satisfaction of creation paired with the drain of relentless mental effort.
I don’t think traditional burnout frameworks capture this. Burnout usually implies you’re doing something you don’t want to do. I want to do this. I love doing this. But doing it continuously, without natural breaks, is wearing me down in ways I didn’t anticipate.
The Practical Reality: What Keeping Up Actually Requires
Here’s what maintaining this pace actually takes:
Continuous high-level thinking. You’re not writing boilerplate anymore. You’re making architecture decisions, reviewing generated code, catching edge cases, explaining business logic. This is mentally demanding work. And it doesn’t stop, because the agent doesn’t get tired.
Discipline to pause. It’s 11pm. I should be asleep. But I could implement one more feature. The agent is there. The conversation is flowing. Stopping feels artificial when progress is possible. Before, you’d hit a blocker and be forced to stop. Now, you have to choose to stop. That’s harder than it sounds.
New patterns for sustainable work. Traditional development had natural rhythms: research phase, implementation phase, debugging phase, deployment. With AI assistance, these compress. You can research, implement, and deploy in an afternoon. The old patterns don’t work. New ones haven’t fully emerged yet.
Managing the compulsion. Because you can build that feature tonight, the temptation to do it is strong. The barrier between idea and implementation is so low that deciding not to build something requires active restraint.
This isn’t a complaint. This is the actual experience of working at 10x productivity. It’s amazing. And it requires different energy management than traditional development.
What I’m Not Ready to Give Up
Despite the fatigue, despite the relentlessness, despite the uncertainty - I’m not ready to stop.
The ability to turn ideas into reality quickly is intoxicating. Not in the addiction sense (though maybe that’s part of it). In the “this is what I’ve wanted my entire career” sense.
I wanted to build things that matter. I wanted to solve real problems. I wanted to move fast. AI-assisted development makes that possible in ways that weren’t before.
Yes, it’s exhausting. Yes, it’s relentless. Yes, I probably need better boundaries. But the alternative - going back to filing ideas away in the “someday” folder - feels worse.
What I’m Learning About Sustainable Pace
Three months of 10x productivity has taught me some things:
1. You can’t sprint forever, even when it feels effortless. The creation feels easy. The mental load isn’t. You need rest even when you’re not physically tired.
2. New productivity requires new boundaries. “I’ll stop when I hit a blocker” doesn’t work when blockers are rare. You need different stopping points. Time-based? Feature-based? Energy-based? Still figuring it out.
3. The possibilities are worth the effort. Yes, it’s demanding. But I’ve built five production applications in three months. They work. People use them. That’s extraordinary. The fatigue is real, but so is the accomplishment.
4. We’re all learning this together. There’s no playbook for sustainable 10x productivity. Yegge talks about 4-hour workdays. Maybe. Or maybe we find new rhythms that work. The point is: we’re figuring it out as we go.
The Honest Assessment
AI-assisted development is genuinely transformative. Not hype. Not speculation. Actually, measurably, right-now transformative. I have five production applications to prove it.
The practical reality of keeping up with that transformation? It requires:
- Sustained high-level thinking
- New discipline around when to stop
- Different energy management than traditional development
- Acceptance that “always-on possibility” needs boundaries
Is it worth it? Absolutely. I’ve built things I’ve wanted to build for a decade. Ideas don’t die in the “someday” folder anymore. The creative satisfaction is real. The productivity gain is real.
Is it demanding? Also yes. The mental load of continuous decision-making is real. The temptation to never stop is real. The need for new sustainable patterns is real.
Both things are true. The possibilities are enormous. And keeping up with those possibilities takes work.
What’s Next
I’m not slowing down. Not because I can’t, but because I don’t want to. The ability to turn ideas into reality quickly is too valuable to give up.
But I am learning to pace differently. To recognize when I’m mentally tired even when I’m not physically tired. To stop when I’ve reached capacity, even when more progress is possible. To treat this like a marathon, not a sprint.
The AI vampire drains you not through malice but through possibility. It makes creation so accessible that stopping feels like waste. That’s the practical reality.
And you know what? Even knowing that, even feeling tired, I wouldn’t go back. The decade of ideas sitting in the “someday” folder versus three months of building and shipping? I’ll take the exhaustion and the apps.
Because this isn’t just about being more productive. It’s about finally being able to build the things you’ve always wanted to build. That’s worth figuring out sustainable pace for.
The possibilities are enormous. The challenge is real. And both are worth embracing.
Sources: