A prototype
with a specific job.
An evidence prototype is organized around the behavior to observe and the decision that observation should inform.
03
What does a working slice need to establish?
Choose the slice around the uncertainty.
A prototype can connect a source, a transformation, and a visible result without building every surrounding feature. The useful slice depends on the question: an interaction-only mockup cannot establish the behavior of an untested external dependency.
Start by identifying which part of the path must be real for the observation to matter. Other parts can be simplified when that boundary is clearly documented. The purpose is a focused technical investigation rather than a convincing imitation of a finished product.
Label what is real and what is simulated.
A fixed response, a sample file, or a substitute service can make a trial practical. Each also changes what the trial can establish. Mark these substitutions so reviewers do not treat convenient behavior as evidence about the real dependency.
Potential scope includes a vertical technical proof of concept, limited data-processing prototype, compatibility demonstrator, or a working example for a specific dependency. Production hardening, broad integration, and continuing operation need their own agreed scope.
Design the observation into the build.
The prototype should expose the information needed to answer the question. A visible output may be enough for one trial; another may need a trace of intermediate values or a controlled comparison. Collect relevant evidence without turning the exercise into an unrelated monitoring project.
Plan a result that would challenge the preferred assumption. If every possible outcome can be described as success, the prototype may be demonstrating a narrative rather than testing a claim.
Explain what can travel forward.
An agreed delivery can include the prototype, setup instructions, sample description, observations, and a note on reuse. Some parts may be suitable for a later implementation; others may exist only to make the investigation possible.
The report should distinguish those roles. Prototype code does not become production-ready simply because it answered the original question, and a useful negative result can still provide a sound basis for the next decision.
What the scope
can produce.
- A question-led technical slice
- Visible real-versus-simulated boundaries
- An observation route appropriate to the claim
- A result record with reuse limitations