In celebration of numbers

My mind has been full of numbers lately. April has Tax Day, so I stress over my numbers and am usually disappointed at how they work out.

But I was also made aware of some very good numbers this month – 50,000 and 5 billion. The Tamarack AI™ Product Suite now has more than 50,000 active contracts carrying over $5B in assets under management. When I was a young professional, I always wanted to be a part of a product launch that delivered millions. I am very excited about being part of something that is doing billions.

But not all data is numeric. In fact, I was surprised with how our data team answered the question “What percentage of the data we manage is numeric vs. text?” I have always thought of equipment finance as an accounting-based industry, so I expected the answer to be “90% numbers.” The consensus was that 75% of the data we manage is text. There are a lot of names and descriptions associated with each number.

As is always the case, the answer to my first question created another: “Does AI care about text vs. numbers?” I knew the answer to this question because language is recorded as text and generative AI is built on large language models. ChatGPT loves text. But operational AI, predictors built with machine learning models, uses both. We have predictors that forecast numbers like interest rates, and we have predictors for classifications like Good-Late-Loss delinquency.

AI uses both numbers and text — which is a good thing.