Increased booking conversion by 160%
Reduced call center load by 42%
Enables a 28% reduction in no-shows
Helps clinics save +8 admin hours per week
Increased patient adherence and preparation for their appointment by 35%
Spryt uses Google Cloud to power Asa, an AI scheduling agent that helps patients overcome barriers to care.
Spryt uses Google Cloud to power Asa, an AI scheduling agent that helps patients overcome barriers to care.
Replatforming on Google Cloud for healthcare at scale
When people fail to attend healthcare appointments, the consequences can extend far beyond wasted resources, with patients potentially missing out on critical screening, diagnosis, or care. Non-attendance is often attributed to forgetfulness, but the team at Irish healthtech startup Spryt sees things differently. “The healthcare system traditionally labels a missed appointment as ‘Did Not Attend,’ treating it like a patient failure,” says Daragh Donohoe, CEO of Spryt. “We see it differently. It’s fundamentally a system design problem.”
In the wake of the COVID-19 pandemic, Spryt created Asa, a virtual, AI-enabled agent that helps patients overcome barriers to attending care. “People miss appointments for very human reasons: fear, embarrassment, or even confusion about why the appointment matters,” Donohoe explains. Asa predicts which patients are at risk of non-attendance, engages them in their preferred language through trusted messaging channels such as WhatsApp and SMS, and helps them book, ask questions, and even arrange a ride to the clinic using a transport agent in partnership with MediDrive.
The solution originally ran on a fragmented AI stack that included a conversational framework, a third-party retrieval-augmented generation service, and several language models. The company was in need of a unified platform that could reduce its operational burden, improve performance, and comply with the security and data protection requirements of major healthcare organizations. For this, Spryt replatformed on Google Cloud.
We opted for Google Cloud because it gives us the AI infrastructure of a global enterprise, with healthcare-grade compliance built in.
Daragh Donohoe
Co-founder and CEO, Spryt
“Our team was spending more time running infrastructure than improving the product and we sell into the most security-conscious buyers in the world,” Donohoe explains. “We opted for Google Cloud because it gives us the AI infrastructure of a global enterprise, with healthcare-grade compliance built in.”
Using Gemini to remove barriers between patients and essential care
For an overdue cervical-screening patient, for example, Asa can initiate a WhatsApp conversation, help them book an appointment, and provide reassuring, clinician-approved responses to any questions she may have. Asa is the front of a multi-agent system. Behind it sits Cortex, the Gemini-powered decision layer: before Asa says a word to a patient, Cortex checks it for safety, privacy, compliance, and clinical appropriateness, and decides when a human should take over. Sentinel, a bespoke monitoring model also built with Gemini, then watches the system over time, continuously reviewing conversations for quality, tone, and safety. TensorFlow-based prediction models identify patients at risk of missing or canceling appointments. During our participation in Mayo Clinic Platform_Accelerate, we developed and evaluated the models using more than 224 million de-identified records available through Mayo Clinic Platform. A behavioral and linguistic layer helps tailor each message to the patient’s likely barriers, preferred language, at the moment when support is most likely to make a difference. The platform runs on Google Kubernetes Engine, with Cloud SQL supporting its MySQL database and Gemini Enterprise Agent Platform used to build and train its models.
“Gemini powers Cortex, the decision layer that governs everything Asa says. It was the right fit: multilingual fluency, plus the instruction-following and function calling Asa needs to take real actions in clinical systems,” Neill Dunwoody, CCO says. Gemini Enterprise Agent Platform provides a shared environment for developing, evaluating, and serving the models behind Asa, Cortex, Sentinel, and Spryt’s TensorFlow based prediction systems.
When a health authority puts us through security due diligence, being on Google Cloud keeps the infrastructure conversation short. That enables us, a startup from Monaghan, to compete in rooms that are usually reserved for much bigger companies.
Neill Dunwoody
Co-founder and CCO, Spryt
The healthcare industry’s stringent data privacy and compliance requirements make security just as important as performance.
“Asa was born inside the NHS, one of the most regulated healthcare systems in the world, so rigorous governance is built in, not bolted on. On Google Cloud, patient data is processed in-region, encrypted at rest and in transit by default, and serving models through Gemini Enterprise Agent Platform keeps patient data secure,” Dunwoody says. “When a health authority puts us through security due diligence, being on Google Cloud keeps the infrastructure conversation short. That enables us, a startup from Monaghan, to compete in rooms that are usually reserved for much bigger companies.”
Looking to the future with AI-enabled end-to-end care journeys
In a deployment with the British National Health Service (NHS), Spryt increased booking conversion by 160% and saved each participating practice more than eight hours of administrative time per week. With support from Google’s Startups’ Growth Academy: AI for Health programme, the company was able to develop its internationalization programme and they are in the process of integrating onto the Google Cloud Marketplace. In the U.S., Asa is now deployed across multiple clinics, delivering thousands of appointments each month. Within a single quarter, Spryt observed a 28% reduction in no-shows, a 42% reduction in call-center load and a 35% increase in patient readiness and adherence for their appointments.
Looking ahead, Spryt’s moonshot for AI-enabled healthcare is a shift toward full 'care journey orchestration.' In this model, intelligent agents handle the complex logistics of the medical system, seamlessly coordinating every step of a patient's care itinerary to deliver a truly frictionless experience from start to finish. “Gemini provides the reasoning required to orchestrate whole pathways, not just single appointments. Ultimately, our aim is for Asa to coordinate every appointment across a patient’s care pathway, ensuring what’s due gets booked and what’s booked gets attended,” Donohoe says.
Google Cloud takes the infrastructure, scale, and the security questions off our plate, so a team our size can focus on the only metric that matters: turning scheduled care into delivered care.
Daragh Donohoe
Co-founder and CEO, Spryt
The team is looking forward to continuing to work with Google Cloud, with plans including natural voice capabilities with Chirp for patients who are less comfortable with text-based messaging. “Google Cloud takes the infrastructure, scale, and the security questions off our plate, so a team our size can focus on the only metric that matters: turning scheduled care into delivered care,” Donohoe concludes.
Founded in County Monaghan, Ireland, Spryt is a healthtech company using AI to remove barriers to attendance and help healthcare providers deliver care more efficiently.
Industries: Healthcare and Life Sciences, Startup
Location: Ireland
Products: Google Cloud, Chirp, Cloud SQL, Gemini, Gemini Enterprise Agent Platform, Google Kubernetes Engine
