I’ve been writing software professionally for over 20 years. I started when “version control” meant emailing yourself a zip file of your code. I’ve been through the rise of web applications, mobile, cloud computing, and now AI-assisted development.
Each of these shifts changed how I work. The AI shift is different from the others in an important way: it’s accelerating the pace of development in ways that scale with expertise.
Here’s what that means.
AI Amplifies Experience, It Doesn’t Replace It
The narrative in the press is that AI is democratizing software development — letting anyone build sophisticated software without deep technical knowledge. There’s some truth to that for simple applications. For complex production software, the reality is different.
AI tools are most valuable in the hands of experienced developers. Here’s why:
Experienced developers ask better questions. AI gives you better output when you give it better input. Knowing what to ask — how to describe a problem, what constraints matter, what failure cases to consider — comes from experience.
Experienced developers can evaluate AI output. AI generates code that looks right. Knowing whether it is right requires expertise. A junior developer might miss the subtle bug in AI-generated code. I’ve been doing this long enough to catch it.
Experienced developers know when not to use AI. Security-sensitive code, complex business logic, architectural decisions — these need human judgment. Knowing where to apply AI and where to apply experience is itself a skill.
Experienced developers have context AI doesn’t. After 20 years in production systems, I’ve seen what goes wrong. I build defensively against failure modes that newer developers don’t know to think about. AI doesn’t have production incidents in its training data — it has documentation.
The Speed Differential Is Real
I’ll give you concrete numbers. On a recent project — a custom order management system for a manufacturing client — I tracked my time carefully.
Equivalent work two years ago would have taken approximately 10 weeks of development time. With AI-assisted development, I delivered it in 7 weeks. That’s a 30% reduction in calendar time on a project of real complexity.
Where did the time go?
- Boilerplate and scaffolding: down by ~60%. AI handles this faster than I can type.
- Standard CRUD features: down by ~40%. Pattern-based work is AI’s strength.
- Integration code: down by ~25%. First drafts come fast, but the edge cases take the same time.
- Complex business logic: down by ~10%. This is the thinking work. AI helps me organize my thoughts but doesn’t speed up the judgment.
- Testing: down by ~30%. Test stubs generate fast; edge case tests still take time.
Accumulated over a project, this is meaningful. More delivered in less time.
What This Means for Clients
Faster delivery. Projects that would have taken 3 months in 2022 take 2 months now. That’s real business value — faster time to market, faster ROI on the investment.
Same quality floor. Faster doesn’t mean sloppier. The quality standards are the same — proper testing, secure code, maintainable architecture. The speed improvement comes from automation of the repetitive parts, not from cutting corners on the important parts.
Competitive pricing. When I can deliver more in less time, I can price projects more competitively. I pass some of the efficiency gains along to clients.
The Limitations I’m Honest About
AI tools have made me faster. They haven’t made me omniscient. Some things are still hard:
- Understanding a new client’s business well enough to build software for it takes time and conversation
- Complex integrations with legacy systems are still research-intensive
- Performance debugging in production still requires methodical investigation
- The hardest architectural decisions still require sustained human thinking
And one genuine risk: there’s pressure to move faster than the quality floor should allow. I use AI to move faster within my quality standards, not to lower my quality standards. This requires discipline.
The Bottom Line for Your Project
If you’re hiring a developer or development company for custom software, AI capability matters. But it’s not the capability itself that matters — it’s whether the person using AI has the experience to use it responsibly.
Two developers with the same AI tools produce very different outputs based on their experience, judgment, and understanding of your domain. AI amplifies what’s already there.
I have 20+ years of experience building software for real businesses. AI has made me faster. That combination is what I offer.
Let’s talk about what you’re building and whether I’m the right developer for it.