My AI development workflow is not perfect, and it will keep changing. What matters is the structure: define the project, build the guardrails, work in small phases, and stop at manual gates.
This is for larger, more complex software. You still need to understand Git, testing, and the basics of software development. One giant prompt that loops until an agent declares victory may produce a demo, but it is not how I build software I plan to maintain.
read moreAI lets me build workflows and environments I have wanted for years. I can finish in hours what used to take days, explore an idea without sinking a month into it, and review more code than I ever could alone.
That does not mean AI has solved software development. It has moved the bottleneck. Code is becoming cheap, but good ideas, clear requirements, architecture, communication, and judgment are not.
read moreInstead of guessing what AI might become, let’s look at what it can actually do today. These are the projects and workflows I have completed with AI as of July 2026.
The short version is that AI is useful, but it is not magic. It is excellent at research, documentation, code review, scaffolding, testing, and shortening long debugging sessions. It is far less reliable when someone asks it to build complex software without understanding the system underneath.
read more