The AI-based attribute tagging is a valuable feature. It passes through text data and identifies the tag-words and keywords and connects them to various attributes in the whole system. The system was supposed to run through a lot of existing data in terms of which tag-words would reflect which keywords. There was a model built on top of that. We were building a machine-learning model, which passed through all of the data and did the necessary attribute tagging. We couldn't find attribute tagging in other services. We initially tried to do it in-house, but we couldn't get the accuracy that we wanted. Elasticsearch was quite efficient in terms of getting accuracy with the limited amount of data that we had. We had 10,000 to 20,000 records. Based on that, we had a good amount of accuracy, which we were happy with. There's a lot we can do with customization.