Organizations often struggle with finding nuggets of information buried within their data to achieve their business goals. Technology sometimes comes along to offer some interesting solutions that can bridge that gap for teams that practice good data management hygiene. We’re going to take a look deep into the recesses of creativity and peek at two interesting possibilities in using virtual reality and augmented reality (VR and AR) as well as artificial intelligence (AI) to enable businesses to interrogate data in new and interesting ways.
Virtual Reality (VR) and Augmented Reality (AR)
Back a few years ago, a friend and colleague of mine were working at a manufacturing company. We were in the midst of replacing a warehouse management system that was focused on optimizing picking routes for stock pickers. Essentially, a software solution that knows the quantity and location of all parts within the warehouse — as well as when the manufacturing floor needs parts as part of the process of making goods — the software would chart the most optimal path through the warehouse for the stock picker to travel as they collected the items, thus getting materials up to the floor as fast as reasonably practicable.
We had joked with each other, asking: “If time and budget were no factor, what would we build?” That answer became really interesting. Fueled by the summer of “Pokémon Go,” we had thought it would be interesting to look at the nature of our warehouse and demand from the production floor with the perspective of augmented reality, highlighting the path in a stock pickers device. We then went off the deep end of creativity: “What if we could move this to virtual reality and show some data?”
We had postulated that, armed with a VR headset, a business user could see the value of inventory in the warehouse and the length of time it takes to turnover, and then calculate the cost of storage. We would also be able to visualize the total amount of end-product that the warehouse floor inventory could produce. Suddenly, we could visualize the amount of lead time our warehouse could provide in terms of actual physical products — rendered in a tangible, life-sized VR environment.
This turned out to be a fool’s errand at the time, and a silly idea that would never come to pass. And so, we opted for a more traditional solution and implemented a fairly standard warehouse management system. However, technology today is allowing something like this to be a feasible solution. With tools like Microsoft HoloLens and Apple Vision Pro, the technology landscape is starting to support our crazy ideas. Taking a simple and standard warehouse management system and using an algorithm for best path optimization and visualizing that with these new AR tools seems a simple feat. Heck, even some of our VR and AR dashboard ideas don’t seem that far-fetched anymore.
If implemented as a VR solution, it would allow for an upper-level manager to review the state of a warehouse without having to necessarily step foot there, perhaps uncovering and visualizing that state in terms of tangible things. Seeing a number on a spreadsheet or dashboard that represents the outcome of Material Resource Planning (MRP) versus seeing the set of products that can be built and leftover inventory after that build is rather illuminating.
Artificial Intelligence and Conversational Data Analysis
With a lot of ballyhoo around AI these days, it’s tough to separate the wheat from the chaff. A few years ago, at an Enterprise Data World event, I had the pleasure of listening to Bill Inmon talk about “textual disambiguation,” as he called it at the time. With the nature of large language models (LLMs) and what we can collect and train LLMs on, the nature of Bill’s genius idea takes on a bit of new meaning.
Often sorting through and understanding customer feedback can be challenging. Organizations have sometimes used word clouds and other visualization concepts to decipher meaning and feedback from unstructured, user-entered text. As we often know, star ratings and the like don’t tell a complete story. What Inmon was proposing was around using natural language processing to derive meaning from text and use that process as part of an ETL (Extract Transform Load) to populate a data warehouse for reporting.
Now, we can train an LLM on a high volume of customer feedback and navigate the results conversationally with AI, as well as meet the end goal of textual disambiguation by use of an API (Application Programming Interface) to that LLM. This allows a business user to prompt the LLM with questions like, “Do customers have a positive experience using our product?” The LLM can determine when negative words are used with a positive connotation and provide accurate results, whereas a word cloud does not provide such insight. Furthermore, a business user can ask more detailed questions of the LLM, such as, “What is the customer experience in using the red safety guard on our newest mandolin slicer?” The LLM can quickly understand and aggregate all customer feedback that mentions such a feature and provide an answer quickly.
This conversational way to explore data provides a very business-friendly way to do meaningful data analysis in a way that hasn’t really been done before.
