Workflow & practice

Prototyping

Prototyping is using an agent to throw together a rough, disposable version of something fast, so you can see an idea working and decide what to actually build. You optimise for speed and learning, not polish, and you discard the result freely.

James Phoenix
Understanding Data Updated July 2, 2026

Prototyping is using an agent to throw together a rough version of something fast, so you can see it, poke at it, and decide what to actually build. The output is disposable by design. You are buying information about the idea, not shipping the code. Prototyping is where vibe coding earns its keep: you steer on feel because the code is not meant to last.

Why agents changed this

Prototyping used to carry a cost: even a throwaway version took real hours, so you thought twice before building one. Agents collapse that cost. You can stand up three different approaches in the time it once took to argue about one, then keep the version that feels right and bin the rest. The cheapest way to learn whether an idea works is often to build a scrappy version and look at it.

The trap is letting the prototype quietly become the product. Code written for speed skips the things that make software safe to run: error handling, tests, structure. That is fine while it is a prototype and a real problem the moment it is load-bearing.

  • Optimise for speed and learning, not correctness or polish.
  • Discard freely. A prototype that taught you something did its job.
  • If you decide to keep it, rebuild it properly rather than patching the throwaway.
Tip
Prototype first, then write the spec or design doc. A working rough version is the best possible input to a real plan, because it turns vague opinions into something concrete you can react to.

Related terms

Building with AI agents?

This dictionary is part of how I think about agentic engineering. If you want the same thinking applied to your codebase, that is what I do.

See how I can help