Imagine going to the ER for a serious health issue. The doctor gives you important follow-up instructions, but once you're home, you're unsure what to do next. You need to schedule a specialist appointment, refill a prescription, or track symptoms—but navigating the system feels overwhelming.

For millions of patients, this is the reality of modern healthcare.

Missed checkups lead to worsening conditions. Scheduling a doctor can mean hours on hold.

Rely Health is fixing this.

They equip care navigators with the latest AI technology to deliver high-quality and cost-effective support—so you get the proactive care you deserve.

But building these AI solutions came with huge technical hurdles:

• Updating prompts was risky. Small changes to one AI workflow could break another, leading to unpredictable errors.
• Debugging was slow and painful. If something went wrong, engineers had to manually test one prompt at a time to find the issue.
• Deploying AI workflows took months. A single new workflow required extensive coding, slowing down innovation.

The stakes were high.

“Patients are bouncing back to the emergency with worsening symptoms because they are unable to navigate the labyrinth of healthcare on their own. Rely Health’s Care Navigators, empowered with Agentic AI, are changing this.” – Dale Ellicott, VP of Growth at Rely Health

To achieve this, Rely Health needed a better way to iterate on AI prompts, track their impact, and scale patient care without engineering bottlenecks. So they found a solution with Vellum.

What once took dozen engineers up to eight months can now be built in 24 hours by a single engineer using Vellum.

We sat down with the team behind Rely Health to hear their story. What follows is a behind-the-scenes look at how they build and evaluate their AI solutions — so you can get the care you need, exactly when you need it.

AI development, at scale

Rely Health’s mission is simple: make sure every patient gets the right care at the right time. But that looks different for every hospital and doctor, which means AI needs to be flexible.

That’s where their Content Engineering team comes in. They work alongside care navigators to understand what parts of their job feel repetitive or manual—and then find ways to automate them.

Today, RelyHealth offers various AI solutions:

• Voice agents: Each client gets a list of customized AI-powered phone workflow. These voice agents automate tens of thousands of routine, lower acuity calls for navigation teams (who only get looped in when a patient needs additional help). RelyHealth customizes these calls to the precise needs of the client and navigators workflow. These agents can handle after-hours patient calls every day!
• Patient check-ins – Automated follow-ups to keep care on track.
• Smart triaging – Screens patients and seamlessly hands them off to human navigators.
• Automated charting—Transform clinician-patient interactions into structured EHR data automatically. Additionally, they autonomously identify actionable insights and next steps for their navigators and care teams.

"We create dozens of AI workflows—easily 7-8 per client, covering patient phone calls, provider calls, and more. Vellum has been a huge help because we need to rapidly iterate and deploy. We can take a given execution, drop it into our evaluation table, and keep moving fast." says Prithvi, CTO at Rely Health.

All of these are done with less engineering effort and no delays.

No more engineering delays

Before Vellum, building AI workflows meant writing code for every step and manually testing each one—slow, tedious, and tough to scale. Every change required more engineering time, making iteration a challenge.

With Vellum’s visual Workflow Builder, Rely Health’s team can design, tweak, and deploy AI workflows in a fraction of the time. Just drag, drop, and refine until it works.

"Before Vellum, we had to manually build every workflow from scratch. Now, a single content engineer can do what used to take a dozen engineers months. We use the Workflow Builder to test what’s possible and scale once we prove it works." - Prithvi Narasimhan, CTO & Co-Founder at Rely Health told us.

Now, they experiment freely and scale with confidence.

"It’s not just about speed—it’s about being able to adapt fast. If something isn’t working, we fix it right away instead of waiting weeks for engineering." added Natoli, a content engineer in their team.

And that speed doesn’t stop at building workflows—it extends to testing them, too.

Catching Issues Early

With Vellum’s Evaluation Suite, Rely Health automatically test hundreds cases at once, making iteration faster and AI workflows more reliable.

"Before, we had to check every single case manually, which made iteration slow," Natoli said. "Now, we run tests in bulk, spot issues instantly, and improve AI performance without the guesswork."

With automated evaluations, they make sure their AI is asking the right questions, making the right decisions, and continuously improving—before it even goes live. And because Vellum’s Eval product is customizable and intuitive, they can shape it to fit their exact needs.

"The platform works exactly how we need it to." Prithvi adds.

Debugging In Minutes

Before, if an AI workflow failed, engineers had to manually test each prompt to find the problem. Changing a prompt risked breaking something else. Now, their team uses Vellum’s Observability tools to capture executions in production, replay them and update safely.

“When one of our apps faces an issue in production, a doctor would contact us, and we’d resolve it in under 10 minutes. Vellum’s tracing views made it easy to identify exactly where the problem is," said Natoli.

With Vellum’s observability tools, they can trace errors instantly, fix them, and deploy the update—all in minutes.

AI That Supports 100% of Patients

"A case manager can only handle 10% of patients. AI can help all of them." - Prithvi Narasimhan, CTO & Co-Founder at Rely Health

There’s still a long way to go, but Rely Health’s AI-powered workflows are already transforming patient engagement:

• Doctors’ follow-up times cut by 50% – AI navigators handle initial outreach and triaging.
• Care navigators now serve all patients, not just the most critical 10%. As a result, this helps reduce hospital readmissions, shorten length of stay, and improve NPS.
• Faster, safer AI updates – New workflows and prompt changes go live instantly without disrupting existing cases.

"Having a platform we can lean on is critical for patient care. We’re so glad we found Vellum to be our partner in this mission," said Prithvi.

At Vellum, we’re proud to support a mission that makes real impact—helping Rely Health bring better, faster care to more people.

Build AI Systems You Can Trust

For Rely Health, AI isn’t just about automation—it’s about patient care. By combining human expertise with AI-driven workflows, they’ve ensured no patient is left behind.

With Vellum, they can confidently scale, refine, and deploy AI-driven healthcare solutions—knowing they’ll work reliably in production.

If you want your team to build faster and deploy reliable AI systems with confidence, schedule a demo with Vellum.