AI for Retail Design: Beware the Shiny Tool(s)!
with new tools and models coming out every month, beware of the shiny tools... now more than ever.
AI is cool, that goes without saying. The challenge is how can you separate out what looks cool versus what makes your company or brand better? When something it’s just a shiny tool vs a game changer for your brand.
In simple terms, ‘better’ can be defined as whether the solution is helping provide better information to make more informed decisions and/or reducing workload. From a technology perspective, ‘better’ could also mean whether the solution will allow your users to pivot easily and quickly to leverage new AI capabilities and whether that data can easily be shared with the extended enterprise.
Moore’s law is a well-known principle stating that computer capacity and processes will double every two years. This principle can be applied to AI except we’d say the timeframe is every two months as opposed to every two years!
Key concepts to think about for a solid ai implementation strategy
You don’t need to have everything figured out before you can leverage AI, but you should ask the question to assess its impact to your business:
- Is it just a simple tool that meets the needs of one user group? If it does, how can the information in that tool be leveraged and/or shared with other groups? For example, if the tool is being used by design, how does the information in it get shared with the product developers?
- Does the tool provide a way to easily consume data from other teams that the one group needs? For example, if the tool is being used by design to create images, how does it leverage existing assets such as prints created by the print design team? How can this information be shared with merchandise planning tools?
- Is the tool installed on the desktop? Desktop applications require more investment to set up and keep updated as opposed to a web-based solution. Web based solutions can scale better without manual intervention and hardware upgrades.
- Does the tool support multiple AI models? Each AI tool has different strengths and weaknesses so the ideal solution leverages some or all of them! If it only supports a single model, then you may miss out on new breakthroughs.
There are specific scenarios when isolated tools make perfect sense. For example, if I’m generating images for a PowerPoint, then a simple tool is perfect. However, if I’m generating design images as part of the ideation process to eventually hand over to merchandising and/or development then I need something a little more comprehensive otherwise I’m investing more time and effort to distribute it.
Here’s where the sentence: Beware the shiny tool comes back into the picture. The key is being able to step back from what is a really ‘cool looking’ tool and make sure it aligns with your business needs. Admittedly though, it’s sometimes hard to not get emotionally attached to a good-looking tool that demos extremely well!
