PROMPTS serves patients at public clinics in 24 Kenyan counties according to an NPR report from September 17, 2026. An AI system understands SMS questions from women, ranks them by urgency, and flags potential emergencies for medical staff. Jacaranda Health's technology director Jay Patel told NPR that when he joined, fewer than 100 questions arrived daily; now there are 15,000. The service grew from SMS appointment reminders sent by a network of Kenyan pregnancy clinics. When recipients began replying with questions, the exchange evolved into a question-and-answer conversation. Patel stresses the service runs over SMS, costs nothing for users, and works on basic mobile phones.
How urgent cases reach nurses
A natural language processing pipeline assigns each incoming message to an intent category and priority level. Roughly 86 percent of cases receive automated responses; the rest go to trained clinical staff at a round-the-clock helpdesk who answer high-priority cases within a minute. Patel explains that messages flagged as potentially urgent bypass all AI functions and land in a separate queue where the help team can review the full conversation history and call the mother directly. The system flags roughly 7 percent of messages as potentially urgent.
Why fast help matters in Kenya
Kenya loses roughly 6,000 women annually to pregnancy or childbirth complications, about 16 women daily. Nearly one-third of these deaths link to delays in seeking medical help, NPR reports. Lisa Mushega, policy expert at health provider alliance HENNET, told NPR that direct access to a specialist prevents such delays; Jacaranda Health is itself a HENNET member. Mushega describes women who recognized signs of preeclampsia through PROMPTS and another mother whose newborn was severely ill. The latter was advised to seek the next facility for her baby's jaundice.
What controlled studies show and don't
A cluster-randomized trial at 40 health facilities across 8 Kenyan counties with 6,139 enrolled pregnant women examined effects of PROMPTS enrollment starting November 2021. Versus control groups, knowledge index and birth preparation index each scored 0.08 standard deviations higher, routine care-seeking index 0.07, newborn care index 0.09, and postpartum care content index 0.06 standard deviations higher. No statistically significant effect appeared for the emergency sign care-seeking index. Study authors note findings rely on self-reports and the trial was not designed to prove effects on health outcomes. Study co-authors are Jacaranda Health employees.
From Makueni start to regional expansion
For Q3 2024, Jacaranda Health reported reaching 2.7 million mothers and 23 Kenyan partner counties in core programs. By March 2025, the Rockefeller Foundation reported three million women had contacted the service, while NPR in September 2026 estimated 700,000 annual users, a different metric than cumulative totals. A partnership with eHealth Africa, announced June 2024, will test and expand PROMPTS with 25,000 mothers in Nigeria's Kano state.
Where researchers see limits
Javaid Iqbal Sofi, an AI policy researcher at Virginia Tech University, asked NPR what happens if the system misclassifies an urgent message as routine, who notices, and how fast. Smisha Agarwal, director of global digital health at Johns Hopkins University, told NPR such interventions work best where functioning health systems already exist. The NPR report identified rapid handoff to human nurses in emergencies as critical, since medical crises leave little room for machine error. A team of 18 nurses monitors PROMPTS communication round-the-clock and answers questions.