How Julep uses technology
Julep Health builds and runs its own AI tools on servers inside the clinic, for both clinical care and day-to-day administration. As far as we know, it was the first clinic in the country to run a fully on-premises, custom-built system for both kinds of work. This page explains what the tools do, what they do not do, and where cloud services are still used.
Why Julep built its own AI.
For the AI that works alongside the clinicians, Julep did not settle for an off-the-shelf product. The clinic built its own. The clinical AI stack is owned by the clinic, runs on local GPU servers in the building, and was designed for the demands of healthcare, where patient privacy is not negotiable and the cost of getting it wrong is high. Four principles guided the design.
Private by design
Our clinical AI runs on servers inside our own clinic. When it works with your medical records, that processing happens on hardware we own and operate; your chart is not handed to an outside AI company. Your medical information stays where it should: with your doctor.
On-premises GPU servers
The clinic operates its own dedicated GPU servers on site. That gives full control over performance, uptime and security, and removes the per-query costs that make cloud AI expensive for continuous clinical use.
Custom-built and wholly owned
The clinical AI stack was built in-house and belongs entirely to the clinic. No vendor lock-in and no per-seat licensing for the system the clinicians rely on every day. The clinic controls the code, the models and the roadmap.
Cost-effective at scale
Owning the infrastructure avoids the compounding cost of cloud AI subscriptions. The savings go toward affordable care and more time with the doctor, not less.
How it helps with your care.
The system supports the doctor. It does not replace the doctor.
Clinical documentation
The AI helps the doctor draft visit notes, medical summaries and clinical correspondence. Your doctor spends more of the visit listening and less of it typing, without losing thoroughness or accuracy.
Diagnostic support and lab analysis
It helps the clinicians read complex lab panels, spot patterns across dozens of markers, and flag results that deserve a closer look. A second set of eyes on the numbers, with the doctor making the call.
Personalized care plans
By pulling together your full picture, lab results, medical history and lifestyle, it helps the doctor write treatment and wellness plans grounded in your own biology rather than one-size-fits-all guidelines.
Medication and interaction review
The system can quickly cross-reference medications, supplements and lab values to surface possible interactions or contraindications: one more check on every prescription decision.
Patient education
It helps produce clear, plain-language explanations of conditions, lab results and treatment options, so you leave a visit understanding your health rather than puzzling over jargon.
What it does not do
It is workflow support for the care team, not a replacement for the physician. Drafts, summaries and flags go to the doctor, who decides what happens next.
Administrative work, handled quietly.
Operational help that lets the team focus on you.
Scheduling and workflow
The system helps organize appointment scheduling, patient communications and routine administrative tasks, which shortens waits and cuts the paperwork that makes healthcare feel bureaucratic.
Communication and follow-up
From referral letters to follow-up instructions, it helps the team produce accurate, professional clinical correspondence in a fraction of the time, so nothing falls through the cracks.
Owned and operated by Julep Health.
Purpose-built hardware and software, run by the clinic.
Local GPU servers
Clinical AI inference runs on premises, on hardware the clinic owns.
Records stay local
Clinical AI works with patient records on our own network.
Fast responses
Sub-second responses for clinical decision support.
Used freely
Owning the clinical AI means it is used whenever it helps, not rationed by per-query fees.
Where we do, and do not, use cloud AI
Our clinical AI, the system described above, runs on our own on-premises servers. Separately, a few online convenience features use carefully selected third-party AI services over encrypted connections: the chat assistant and symptom helper on this website, our SMS texting assistant, and the short educational explanations of lab tests in our patient portal. These convenience tools do not have access to your medical chart.
Care first, technology second.
The tools exist so the doctor can spend the visit with you. Book a visit, or call the clinic with any question about how they are used.
Primary care · Whole Body Assessment · Telehealth · Not sure where to start?