Company Deep Dive: Tennr

Subscribe | Advertise | AI Startup Database | Feature AI Jobs

TOGETHER WITH TENNR

Company Deep Dive: Tennr

Perspectives from the people building the future of health AI…

We sat down with Trey Holterman, Co-Founder and CEO of Tennr, a healthcare automation company focused on referrals. Tennr builds the connective tissue for intake and referral operations, moving patients from messy orders and benefits checks through qualification, prior authorization, and patient engagement, while keeping senders and receivers in sync.

Trey shared how Tennr builds reliable automation under the hood, where patient engagement fits, and why staying narrowly focused on referral-based care has helped Tennr scale quickly.

Let’s start from the top. What is the origin story behind Tennr?
My mom is a family medicine clinician. She would send patients to specialists or imaging and describe it as sending them into a black hole. Valuable patients were getting lost. We started Tennr to do three simple things for receiving providers: increase completed referrals, reduce preventable denials, and improve team efficiency. That is the whole business.

In plain English, what does Tennr actually solve today?
We are laser-focused on referrals. We take a patient from point A to point B. We ingest charts and orders, run eligibility and benefits, apply payer policy to the clinical context to see if a case qualifies, handle prior authorization, and keep the patient, sender, and receiver aligned. We also support post-discharge routing for health systems so fewer patients leak out of network.

What are you using under the hood, and how do the models fit?
We have always bet on open-source models with our own fine-tuning. Historically, we ran a lot on Mistral and Meta’s Llama, and earlier even a Databricks model. Our internal orchestration lets us hot-swap when a stronger model lands. When OpenAI just released its new open-source models it was like Christmas in the office.

Those open source models plus our data continue to pretty dramatically outperform the best closed source models on Tennr tasks. That means fewer items need human approval, which increases safe automation and reduces queue time.

How do you decide what to automate versus send for review?
We set confidence thresholds with the customer. If you demand something like 99% confidence, you might only automate about 20% of the workload. If you accept something like 97%, you can automate closer to 50%. Some customers keep the bar very high. Others compare us to a human baseline and let us run whenever we are above it. The point is that the customer controls the dial and we show the trade-offs.

How is Tennr sold? Is it modular or a full platform?
It is modular and it stacks into a platform over time. Inside the automation layer we offer documentation management, intake operations, eligibility and benefits, qualification against payer guidelines for prior auths to avoid clawbacks, and patient engagement.

Most customers start with one or two modules and expand. There is also a network and visibility layer for live status. Pricing is volume-based on throughput at each step.

What does implementation look like, and how long does it take?
Generally, 3–5 months, although configuration can technically be done in under a week. The calendar time accounts for internal customer approvals, testing, and safety.

For receiving treatment providers, we often re-plumb some front-end operations. We do it once and do it right because we do not want to interfere with patient flow.

Who is using Tennr today? Who is the ideal customer?
In general, treatment providers that are looking to drive up conversions. Mostly high-volume treatment providers across drugs, devices, diagnostics, therapies, and operational procedures. These organizations live on inbound referrals and are buried in paperwork. Customers often start within a service line and then expand.

Walk me through a real ROI example. Where do teams leave money on the table, and how do you quantify it?
We start simple. A specialty group turned on Intake and moved from about 80% to about 85% conversions in the first month. Then we ran a code-by-code denial analysis and showed the CFO where preventable denials were hiding.

Typically, we track three anchors every month: conversions, denials, and FTE efficiency. We only call it a win when all three trend up and to the right with net revenue and hours saved documented line by line.

What new product areas are you most excited about right now?
Patient engagement. As the cost of thoughtful outreach falls, you can actually nurture patients after the referral, not just move them. Calling is starting to work at scale. We will be partnering with one of the largest calling platforms in the space so those workflows show up inside post-discharge offerings, rather than living in a separate tool. 

Seeing patient engagement actually start to work and drive real outcomes is what gets me fired up for our future.

Outside your lane, what excites you most in healthcare AI?
Honestly–referrals. We are monomaniacally focused on referrals. There is still endless surface area in simply getting people to the right care and keeping them there. Focus wins.

Healthcare AI Guy Summary

What stood out, what’s tricky, and why it matters…

Tennr is tackling one of healthcare’s most operationally painful and financially leaky workflows: referrals. Each year, more than one-third of Americans are referred for specialty care, imaging, equipment, or treatment. Whether it’s routing patients post-discharge, managing incoming orders, or navigating payer requirements, the referral process is riddled with faxes, dropped handoffs, and missed revenue. Tennr’s bet is that the right AI-infused infrastructure can fix that.

Founded in 2021 by Stanford engineering friends, Tennr has taken a systems-level approach, building a custom orchestration engine specifically for referral-based care. The result is a fine-grained automation layer that can interpret messy documentation, flag potential denials, and route referrals cleanly across fragmented environments. They’re not just offering assistive tools; they’re automating the full loop.

And they’re moving quickly. The NYC-based company has tripled revenue in the last two quarters, now processes over 10 million documents and hundreds of thousands of referrals each month, and recently raised a $101M Series C from investors like IVP, a16z, Lightspeed, and more. With $162M raised and a $605M valuation, Tennr’s momentum is hard to ignore. The growth story is very real and they’ve built a great culture and team to match the mission.

What stood out

• Referrals are the product, not the wedge: Most companies treat referral management as a feature. Tennr built the whole business around it, and it shows in how deep they go into payer logic, documentation routing, and conversion analytics. They’re best in class here.

• Customers set the automation bar: Confidence thresholds are customizable. Whether you want 99% precision or human-equivalent performance, Tennr lets you decide where to draw the line and shows how that decision impacts volume and efficiency.

• ROI is clear and trackable: Everything is measured using three anchors: conversions, denials, and FTE efficiency. Tennr shows customers how they can help move those metrics with net revenue and hours saved line by line.

• Pragmatic model orchestration: Tennr isn’t chasing frontier models for the sake of it. They rely on fine-tuned open-source models that outperform closed alternatives on their specific tasks and can hot-swap models as the ecosystem evolves. RaeLM, Tennr’s proprietary vision-language model, is trained on over 100 million anonymized healthcare documents, 2.3 billion data fields, and 8,000+ sets of criteria.

• Patient engagement is the next unlock: As outreach tools become cheaper and more integrated, Tennr sees the opportunity to actually guide patients through the care journey, not just move their paperwork.

What’s tricky

• Referral networks are chaotic by nature: Different EMRs, different intake processes, and a lot of fax-based workflows. Tennr’s advantage is being able to slot into that mess, but it also means every deployment needs tight coordination and strong support.

• The threshold trade-off is delicate: Push automation too far and you risk errors. Stay too conservative and you miss the value. Tennr solves this by letting customers move at their own pace, but tuning that balance isn’t always easy.

• It’s still a manual world: Fax is still the default in many organizations. Even the best software can get slowed down by legacy systems, missing data, and human bottlenecks. Tennr’s approach works best when customers are ready to modernize.

Final thoughts

There are few workflows in healthcare more universally hated (and more consequential) than referral management. Tennr is turning one of the most expensive and failure-prone handoffs in healthcare into a high-leverage automation play, blending AI with strong implementation muscle and a clear product wedge.

They’re not trying to replace humans. They’re trying to give teams their time back. If they keep executing like this, Tennr won’t just be a fast-growing automation company; they’ll be a case study in what focus, infrastructure, and product discipline can unlock in healthcare. They are well on their way to rewiring one of the system’s most broken pipes.