Agile AI and Retail

Retail has always been changing,?and even more since Covid-19.?

While data and data science are being used in retail and some retailers are giving their employees data science education, retail needs a different kind of AI. Because, as?Paige Thomas of Sax Off mentioned at the NRF2023, consumers can change their decisions about retail products every morning.

The AI solutions for retail have to be agile and easy to adapt?to change by different business stakeholders in retail. The business stakeholders include front line workers, store associates, executives as well as the data teams. If the AI solution can't adapt easily, it starts not helping but hurting business and is quickly dropped off of the business process by the users. Decision support when it is not any more supporting, is dropped in insurance submission scoring, machine condition monitoring, as well as in retail. As humans, if we are running a difficult course of actions, we need agility and collaboration.

The agility and ease of use of AI needs to happen in model creation, as well as model deployment, monitoring and update. Especially when you have a lot of models in production, models need to be auditable by different stakeholders, so when (not if) something goes wrong, it is easier to debug the data or models. The more eyes know the data and AI model, the faster the debug.

I am really excited about solutions for Demand Prediction, P&L Prediction and Improvement. ?I have seen an availability increase of 4% when about 1K delivery trucks are loaded every morning based on Machine Learning demand prediction models.?P&L Prediction and Improvement?needs to be handled through profitability, retention, price elasticity and front-line worker behavior prediction models. While data science education helps, use of ML platforms are inevitable to create solutions fast and also have those solutions keep performing, while the consumer behavior changes, ML models need to be updated and your employees (and your data scientists) have a higher attrition rate.