If, today, you messaged PROMPTS about an acute headache, you’d receive a response specific to that type of headache. Simple.
Right? In early Q4, our Helpdesk team manually categorized 30,000 ‘headache-related’ questions from mothers into three types, ranging from mild, moderate, to severe.
This matters because different types of headaches signal different things in pregnancy, from simple cases of dehydration to life-threatening complications like preeclampsia. Reacting to these nuances can be the difference between a mother reaching or not reaching lifesaving care in time.
Headaches are just one example of the ruthless approach we’ve taken to improving the accuracy of all messages we send mothers.
The combination of training our AI to detect nuances and improving its functionality to respond to them faster resulted in some significant wins to close the year; an 83% accuracy on all AI responses, a 15% increase in engagement from mothers, and a 24% reduction in questions answered by human agents, helping them prioritize support to urgent cases.