How should we use rapid prototyping to test new ideas?

Rapid prototyping is a way to move from imagining possibilities to testing them. It applies not only to products, but also to ideas: teams make an idea tangible enough to interrogate its assumptions, learn from what does not work and iterate towards what does.

This approach depends on giving people room to imagine freely and build on each other's ideas, rather than immediately narrowing the discussion to feasibility. Prototyping then brings action to that imaginative work. The purpose is not to avoid every unsuccessful attempt, but to make those attempts useful.

We call this practice intelligent failing: quickly testing assumptions, learning from unsuccessful attempts and refining the approach. A prototype becomes valuable through the learning it enables, rather than simply through whether the first version succeeds.

AI can help teams produce prototype ideas quickly, but it does not replace the people who decide what to explore and which ideas to pursue. Strategists, designers, writers and developers bring different perspectives to the work, using AI to generate possibilities they can interrogate, validate and choose from.

In the work discussed around G42's Supercharged event, this cross-functional collaboration was central to accelerating content creation. The team used AI to rapidly prototype ideas while relying on human imagination to work out how the tools could support something new.

The transferable practice is to bring those perspectives together and try ideas in a form that can be examined. Use AI to support experimentation, then apply human judgement to what works. The tools can accelerate the process, but they should not be expected to determine the whole approach.

A working prototype is a stage in development, not the end of the work. For UST's NavigatorAI, we developed a functional prototype in two weeks, then refined, tested and launched the full experience within six weeks. The tool helps prospects navigate UST's AI expertise through tailored questions and personalised recommendations.

The experience combined generative AI with structured rules, trusted content and UST's proprietary expertise. Publicly available business and audience data helped shape each user's journey, while human oversight kept the experience accurate and relevant. Prompt design and optimisation remained part of the team's work.

NavigatorAI illustrates the distinction between making an idea work initially and preparing a complete experience for use. Rapid development was followed by refinement and testing, with attention to the content, controls and human judgement supporting the experience.