We're building the care infrastructure for the hours between visits.
Healthcare works for the 15-minute appointment. We're solving the 10,000 hours that come after, with AI that listens, adapts, and keeps patients connected to their care.
Founders' Story
Matteo Grassi, 3× founder and clinical psychologist. Built an app to monitor patients with bipolar disorder. The tech worked. Engagement didn't. So he threw out the app and called patients directly with AI. Engagement went from 15% to 85%. The problem was never the protocol, it was the delivery.
Sthita Pragyan Pujari was building AI voice agents before voice AI was a category. Ex-Honeywell, scaled Speech AI at Thumb Technologies. Architected HANA's dual-model system, a reasoning engine that thinks before every call, a voice model that speaks, then led the full migration to open-source, self-hosted infrastructure. No OpenAI. No vendor lock-in.
Together, they stopped building another healthcare app and started building the infrastructure that powers all of them.
"While everyone was building cars, we wanted to build roads."
Our Values
Zero friction for patients
If it requires a download, a login, or a behavior change, it's already dead. We meet patients where they are: on the phone, in a conversation. No app. No portal. No burden.
Safety is not a feature
It's the foundation. Over one million patient interactions with zero critical adverse events. Every conversation is monitored, every escalation protocol is clinic-defined, every risk signal is caught.
Infrastructure, not applications
We don't build point solutions. We build the engagement layer that powers everything from intake to monitoring to adherence. Like electricity for patient communication.
Global by design
Healthcare protocols are universal. Incentives vary. We operate across five countries, three languages, and both public and private systems, adapting to each without rebuilding.
AI fills gaps, not jobs
We don't replace nurses and care teams. We put AI into the places where there's no one: missed calls, unmonitored hours, unasked questions. Everyone stays. Care gets better.
Outcomes over demos
We don't do flashy demos or vibe investing. We do 85% engagement, 96% completion rates, and 31:1 ROI. Boring, functional solutions that actually work. That's what healthcare needs.
How We Got Here
The app that nobody used
Matteo built a mental health monitoring app for patients with bipolar disorder. It collected sleep, activity, and mood data. Technically sound. Clinically useful. Engagement: abysmal.
The phone call that changed everything
What if we just called them? An AI voice agent that reached patients by phone and SMS. Engagement jumped to 85%. The problem was never the clinical protocol, it was the delivery mechanism.
From product to platform
Every clinic had different problems (intake, adherence, monitoring, outreach) but the underlying need was the same: continuous patient engagement. So we stopped building a product and started building infrastructure.
One million interactions later
HANA powers patient engagement across five countries, multiple languages, and every care setting from ADHD diagnostics to palliative care. Our reasoning model learns from every interaction.
Cutting the cord: fully open-source
Sthita led the migration off proprietary APIs entirely, moving HANA to open-source models on our own servers. No OpenAI. Full data sovereignty. Now we're testing on-premise installations for hospitals in Italy and sovereign cloud deployments in the Middle East. The future of healthcare AI is owned, not rented.
The app that nobody used
Matteo built a mental health monitoring app for patients with bipolar disorder. It collected sleep, activity, and mood data. Technically sound. Clinically useful. Engagement: abysmal.
The phone call that changed everything
What if we just called them? An AI voice agent that reached patients by phone and SMS. Engagement jumped to 85%. The problem was never the clinical protocol, it was the delivery mechanism.
From product to platform
Every clinic had different problems (intake, adherence, monitoring, outreach) but the underlying need was the same: continuous patient engagement. So we stopped building a product and started building infrastructure.
One million interactions later
HANA powers patient engagement across five countries, multiple languages, and every care setting from ADHD diagnostics to palliative care. Our reasoning model learns from every interaction.
Cutting the cord: fully open-source
Sthita led the migration off proprietary APIs entirely, moving HANA to open-source models on our own servers. No OpenAI. Full data sovereignty. Now we're testing on-premise installations for hospitals in Italy and sovereign cloud deployments in the Middle East. The future of healthcare AI is owned, not rented.
Open source. Self-hosted. Healthcare AI that you actually own.
Six months ago, we made a decision that changed our trajectory: we moved off OpenAI entirely. We migrated HANA to open-source models running on our own servers. No proprietary API dependency. No patient data leaving your jurisdiction. Full sovereignty.
Open-Source Models
We run Llama 3.1 and other OSS models on our own infrastructure. Our reasoning engine doesn't depend on any single foundation model provider.
Data Sovereignty
Patient data stays where it belongs. Our European servers handle GDPR. Our architecture supports region-locked cloud infrastructure.
On-Premise Installation
For hospitals running mainframes and sovereign clouds, we're testing on-premise deployments. Your infrastructure, our intelligence.
The Data Flywheel
Every interaction makes the model smarter. Over one million patient conversations have trained our reasoning engine on clinical nuance.
"We believe the future of healthcare AI is open, sovereign, and owned by the institutions that serve patients, not locked behind proprietary APIs."
Let's close the engagement gap together
Whether you're a clinic, a platform, or a health system, we'd love to talk.
Book a conversation