Unusual custom systems often begin with a sentence that sounds too broad for a proposal: “We have decades of material nobody can search,” “the exhibit needs to make an invisible process understandable,” or “we know the result we want, but no existing product does it.”
The absence of an off-the-shelf answer does not justify an unlimited build. It makes disciplined scoping more important.
Define the changed reality
Start with what becomes possible when the system exists. Who can do what they cannot do now? What decision becomes faster, safer, clearer, or newly available? This outcome should be stated without naming a technology.
Name the boundaries early
Clarify authorization, privacy, safety, physical environment, users, platforms, deadlines, budget, maintenance tolerance, and the material the system must not expose or alter. Boundaries create the shape of the answer.
Build an evidence map
Inspect what already exists: files, devices, partial code, prior prototypes, vendor systems, undocumented workflows, and the assumptions stakeholders disagree about. Separate facts, hypotheses, unknowns, and risks.
When the brief is ambiguous, the first deliverable should be evidence—not architecture.
Prototype the decisive uncertainty
Do not build the whole interface to discover that the camera cannot see the target or the archive format cannot be parsed. Identify the one technical or human assumption that could invalidate the project and test it first.
Use gates, not momentum
At each stage decide whether to continue, change approach, narrow scope, involve a specialist, or stop. A good prototype can prove that a project should not become larger.
Design the operating life
Define who owns the system, where it runs, how it is backed up, what it costs to maintain, how updates are handled, and what happens when the original builder is unavailable.
A useful first brief
- The problem in ordinary language
- The people and materials involved
- What has already been tried
- The privacy, safety, and access constraints
- The most important unknown
- What success changes
That is enough to begin. A polished technical specification is often the output of discovery, not the prerequisite for it.