Prototyping
also called Spike, Throwaway code
Build the cheapest thing that answers the question.
Producing a rough artifact to test an idea or resolve uncertainty, explicitly not meant to be kept. AI tooling makes this dramatically faster, which is one of its clearest wins.
The discipline is labelling the output as throwaway and actually throwing it away. Prototypes that quietly become production are how you inherit code nobody understands.
The discipline that keeps prototyping honest is deciding up front that it will be thrown away, and saying so. Generated code makes direction changes cheap, which is a genuine strategic advantage — as long as the prototype is not quietly promoted into the maintenance surface.
Three sketches instead of one plan
A UI approach is unclear and a decision is needed.
Build all three roughly, look at them, discard two. The cost is now hours rather than weeks, and the decision is made on evidence instead of on the quality of the diagram.
- reportMeasuring the impact of early-2025 AI on experienced open-source developer productivity ↗
A randomised trial on experienced maintainers’ own repositories in which tasks took 19% longer with the AI tools, against participants’ expectation of a speed-up. A snapshot of one population and tool set.
- paperThe impact of AI on developer productivity: evidence from GitHub Copilot ↗
A controlled experiment: developers with Copilot finished a specific JavaScript HTTP-server task 55.8% faster. One narrow task — do not generalise the number to software work.
Vibe Coding
Building by prompt-and-accept, without reading the code.
Spec-Driven Development
Write the specification first; let the implementation follow.
Design Concept
A rough, rejectable sketch of an approach.
AI-Assisted Development
Day-to-day engineering with a model in the loop.