I'll be brief.
The first part of it.... It starts with being able to train the model. We need to bring in engineers. We also have young people annotating data, meaning strings of sentences—for example, “I'm having a bad day.” The young person can come in and annotate that data. You can also bring in psychologists to look at the overall data.
The main thing in terms of the data storage is that, when you train the AI model very effectively to detect signs of mental health—whether the language is neutral, sad or happy—you don't need to store the data. It's kind of like auto-correct or Grammarly. They're not storing all your emails; they're not storing your messages. Essentially, it's kind of filtering through the system, and it's able to get a score. On our end, what we have is a score on—