AI prototypes make Data Products tangible

An under-looked aspect of Data Products in general and AI products in particular is that they are often 'invisible'. They lack a tangibility - something to touch, something to see. This is an unrecognised barrier to good Product decision-making. Reasoning about the invisible is difficult. It's imprecise.

Prototypes make the intangible real. 

They help with product decision-making, with communication, with building stakeholder buy-in, and with customer sales.

Invisibility

AI products often take the form of automating a process, removing something concrete and abstracting it into the automated ether. For customer service, this might mean removing a human from the customer care user journey. For email marketing it might mean replacing the need to manually set up an email campaign with a triggered system. In payments, it might mean removing explicit payment decisions and allowing your fridge to re-order milk when it runs out.

In other cases, the AI product might simply be an optimisation whose benefits are only observed in the aggregate. For example, customer segmentation that drives more effective advertising. Or more relevant product recommendations that drive clicks and ultimately sales. 

Injecting artificial tangibility. 

With these examples in mind, it is clear that invisibility adds a layer of complexity to the already tricky act of product decision-making.

More so, when no single individual has the complete picture of the product and its integrations. Complex product development environments, with many touch-points, moving parts and integrations are breeding grounds for confusion; something that is exacerbated when a concrete understanding of the product at hand is lacking.

A prototype gives a focal point around which discussions can centre. 

This is true, even for exclusively internal stakeholders. On three separate occasions in the past two years, investing time in building a throw-away prototype has been the verifiable differentiator in accelerating progress and securing funding for AI initiatives for my team.

As humans we are well equipped to run simulations and scenarios in our head but this is an exercise that benefits greatly by pinning down some of the uncertainty. Injecting tangibility to the conversation augments our reasoning. It harnesses our ability of thinking by making. 

Importantly, AI prototypes help stakeholders who are less familiar with AI to better understand what exactly the product does. It forces feedback on potential obstacles and accelerates buy-in (or rejection) from customers and gate-keepers. 

AI is still new territory for most. Grounding conversations in the tangible is an unreasonably effective tool for avoiding the twin perils of 'AI can do it all' and 'AI is too complicated'.