When code can be written in an afternoon, the hard part is no longer implementation. The hard part is figuring out what problem to solve, what to build, and when to change course. Implementation details are cheap to change; system design, API contracts, dependencies, and architectural tradeoffs are not. Spend your energy on those durable decisions.
Use the agent as a brainstorming partner during the design phase: have it research options, explore alternatives, and poke holes in your thinking. When two designs are both plausible, have the agent prototype both and compare them. Choices that used to be settled by debate can now be settled by evidence.
Once you’ve committed to a direction, practice spec-driven development: collaborate with the agent to write clear specs and requirements that reduce ambiguity before it starts generating code. An agent given a vague prompt will make its own decisions about tradeoffs, and you’ll spend more time undoing those choices than you saved by skipping the spec.
As you iterate, keep your attention on what matters most, such as whether the architecture will hold up under real load, whether you need to re-evaluate the design tradeoffs you made, and whether the thing you’re building is ready to put in users’ hands.
When code can be written in an afternoon, the hard part is no longer implementation. The hard part is figuring out what problem to solve, what to build, and when to change course. Implementation details are cheap to change; system design, API contracts, dependencies, and architectural tradeoffs are not. Spend your energy on those durable decisions.
Use the agent as a brainstorming partner during the design phase: have it research options, explore alternatives, and poke holes in your thinking. When two designs are both plausible, have the agent prototype both and compare them. Choices that used to be settled by debate can now be settled by evidence.
Once you’ve committed to a direction, practice spec-driven development: collaborate with the agent to write clear specs and requirements that reduce ambiguity before it starts generating code. An agent given a vague prompt will make its own decisions about tradeoffs, and you’ll spend more time undoing those choices than you saved by skipping the spec.
As you iterate, keep your attention on what matters most, such as whether the architecture will hold up under real load, whether you need to re-evaluate the design tradeoffs you made, and whether the thing you’re building is ready to put in users’ hands.